RE Comments on the Long Term 2 Enhanced Surface by ing15204


									                                                                           Government Affairs Office
                                                                           1300 Eye Street NW
                                                                           Suite 701W
                                                                           Washington, DC 20005
                                                                           T 202.628.8303
                                                                           F 202.628.2846

The Authoritative Resource on Safe Water
                                           SM                              Headquarters Office
                                                                           6666 W. Quincy Avenue
                                                                           Denver CO 80235
                                                                           T 303.794.7711
                                                                           F 303.347.0804

      May 18, 2009

      Michael Finn
      U.S. EPA Office of Ground Water and Drinking Water
      Ariel Rios Building
      1200 Pennsylvania Avenue, N. W.
      Mail Code: 4606M
      Washington, DC 20460

      RE:       Comments on the “Long Term 2 Enhanced Surface Water Treatment Rule
                Toolbox Guidance Manual” (Review Draft, February 2009)

      Dear Mr. Finn,

      The American Water Works Association (AWWA) appreciates the opportunity to submit
      comments on the U.S. Environmental Protection Agency’s (EPA’s) “Long Term 2
      Enhanced Surface Water Treatment Rule Toolbox Guidance Manual” (Review Draft,
      February 2009). AWWA hopes that you will find our review of this draft guidance

      This guidance is particularly important to the implementation of the Long Term 2
      Enhanced Surface Water Treatment Rule (LT2ESWTR). The “microbial toolbox” (see
      40 CFR 141.715) was a central concept underlying the Stage 2 DBPR Agreement in
      Principle (AIP), on which the LT2ESWTR is based. The flexibility in the “microbial
      toolbox” that was agreed upon in the AIP was, in theory, to balance some of the
      variability with the analytical method for Cryptosporidium, as well as some other
      assumptions underlying the rulemaking. This Review Draft does not meet the spirit or
      intent of the AIP.

      That agreement and the subsequent Economic Analysis underlying the LT2ESWTR is
      premised on each of the tools included in the microbial toolbox being available for
      compliance. Each one of these tools is important for utilities so that multiple choices are
      available for compliance. In finalizing the LT2ESWTR, the agency set certain “design
      and implementation criteria” in regulatory language (see 40 CFR 141.715(b)). It appears
      to us that through this guidance document, EPA is further constraining the tools available
      to drinking water systems attempting to comply with LT2ESWTR.

      Unfortunately, the language and tone of the draft guidance manual:
Michael Finn
April 17, 2009
Page 2

      1. Describes the need for actions not reflected in 40 CFR 141.715(b) –
         141.720, and
      2. Imparts skepticism that particular tools allowed under LT2ESWTR might
         actually be employed for compliance purposes.

In effect, the guidance leans too strongly toward state primacy agencies only being
satisfied with system compliance through adding additional disinfection to a water
treatment plant. In summary, AWWA believes that in order to finalize the document, the
document should be revised to:

      1. Re-draft existing text on demonstration of performance and riverbank
         filtration per the detailed comments attached; and
      2. Insert additional information pertinent to efficient application of ozone
         treatment per detailed comments which AWWA will forward in the near
      3. Review and delete all uses of the words “must” and “should” that are not
         directly drawn from the regulatory text of LT2ESWTR.

Attached to this letter are detailed comments indicating specific concerns with the draft
guidance document. This attachment is the first of two sets of comments that AWWA
will forward to EPA with detailed recommendations to assist in further refining the draft
guidance. A second set of detailed comments specific to ozone treatment will also be
forwarded to EPA.

AWWA is deeply concerned that this guidance manual has not received any meaningful
attention since the rule was published over three years ago. Now, states and water
systems have actively engaged in rule implementation and this guidance continues to
have significant inadequacies, including issues that AWWA brought to the agency’s
attention at rule proposal. AWWA will appreciate the agency’s consideration of our
concerns and recommendations. If there are any questions, please direct them to me or
Steve Via at (202) 326-6130.

Best regards,

Thomas W. Curtis
Deputy Executive Director
AWWA Government Affairs

cc:      Ron Bergman, EPA/OW/OGWDW
         Pam Barr, EPA/OW/OGWDW
                          American Water Works Association
                                   Comments on the
              “Long Term 2 Enhanced Surface Water Treatment Rule Toolbox
                                  Guidance Manual”
                              (Review Draft, February 2009)

The following contains comments relevant to each of the respective chapters of the “Long Term 2
Enhanced Surface Water Treatment Rule Toolbox Guidance Manual”:

       Watershed Control Program
       Bank Filtration
       Combined and Individual Filter Performance
       Second Stage Filtration
       Demonstration of Performance
       Ultraviolet Light
Introduction [Chapter 1]
The guidance manual does not, but should, communicate the concept of incremental removal credit
under Long Term 2 Enhanced Surface Water Treatment Rule (LT2ESWTR) and previous rules.
Inclusion of text and an accompanying exhibit in the introduction would be most effective. Current
compliance is based on CT and many water system, states, and consultants have developed
spreadsheets for CT compliance. This guidance should describe a framework for comprehensively
describing a facility’s overall removal credit from various activities from the toolbox. Such a
common reference, were it available to water systems, would be a common point-of-departure for
water treatment plant (WTP)-by-WTP discussion of removal credits. This type of framework also
sets the stage for maintaining compliance when a particular unit operation fails to perform (e.g.,
filtration performance criteria are not met); with a summary table of removal at the WTP in hand,
one could easily determine if a system is still in compliance, by looking at the remaining credits
achieved thorough other system activities or unit operations.

Watershed Control Program [Chapter 2]
Chapter 2, Watershed Control Program, is improved from previous draft. AWWA appreciates the
Agency’s considerations of our earlier comments of February 2006. Water utilities will find this
chapter better organized and easier to follow than the previous draft. Several additional comments

   1. Section, states the disadvantages of watershed control program option in such a
      way that states are unlikely to be open to water systems using the tool to comply with
      LT2ESWTR. The tone in combination with the overall requirements imposed, will make
      it unlikely that very many communities will employ watershed control programs for
      purposes of LT2ESWTR compliance.
          “Most water systems who have developed and implemented SWP efforts like
          those needed for the State-approved WCP credit have found that these efforts
      are able to substantially improve source water quality (Ashendorff et al. 1997,
      Vaux 2000). However, seldom will watershed activities result in immediate
      realization of benefits. Many land use policies, wildlife management, and
      public education programs require significant implementation timeframes. This
      challenge is further complicated by the target organism in this rulemaking,
      Cryptosporidium. Cryptosporidium occurs in low concentrations and is
      difficult to detect using existing analytical methods, consequently, it can be
      hard to discern reductions in Cryptosporidium concentrations resulting
      from watershed control programs even if substantial changes are realized in
      the watershed
      Furthermore, it may not be possible to discern the improvements in source
      water quality using these monitoring approaches due to natural environmental
      variability, the characteristics of the source water improvements, and the
      limitations of the current analytical methods for Cryptosporidium or other fecal
2. In Section, Annual Status Report, the following state implies that
   Cryptosporidium monitoring should be an element of watershed control programs under
       "The original watershed control program plan should include specific measures
       by which the PWS can evaluate the effectiveness of the program.” It may be
       helpful to provide examples of how program effectiveness could be evaluated
       in the absence of monitoring.
3. In Section, “Maintenance of the WCP Credit,” the third paragraph references
   “annual watershed sanitary surveys”. This should be changed to reflect that sanitary
   surveys are required only every three to five years.
4. Section 2.3.4 under Watershed Hydrology, the term “vulnerability analysis” is used. The
   term should be changed here, and elsewhere in the guidance, to susceptibility analysis or
   some other term so that it is not confused with terrorism-related “vulnerability analysis”.
5. While Section does explicitly state that Cryptosporidium monitoring is not
   required language in the section and other portions of Chapter 2 imply that
   Cryptosporidium monitoring should be an element of watershed control programs under
       “Influence of Precipitation
       PWSs may find it prudent to determine the extent to which Cryptosporidium
       occurrence in their watershed coincides with extreme rainfall. 68 percent of
       waterborne disease outbreaks between 1948 and 1994 were shown to be
       associated with heavy precipitation (Curriero et al. 2001). Cryptosporidium
       occurrence may also be related to seasonal variations in infection among
       livestock, but any correlation is site-specific and depends on the source.

       Once you have gathered information about Cryptosporidium sources and the
       likelihood of the oocysts reaching your source water (based on watershed
       characteristics and fate and transport), you should determine the amount and
       proportion of oocysts that each source is expected to contribute to the overall

       Cryptosporidium load. Loading can be calculated fairly easily for constant point
       sources such as wastewater treatment plants but is more difficult for farms and
       urban runoff; monitoring and water quality modeling may be necessary (see
       section below on monitoring). What Role Should Monitoring Play in the Evaluation of Potential and
       Existing Sources of Cryptosporidium?
       Monitoring of Cryptosporidium is not required to develop the WCP plan or to
       implement it once approved by the State. The high cost and limitations of current
       analytical methodology pose significant challenges to monitoring for
       Cryptosporidium. In addition, it is noted that it may take years to realize
       measurable improvements in water quality after initiating source water protection
       efforts. Furthermore, discerning improvements in source water quality using
       monitoring can be difficult due to natural environmental variability, the nature of
       the source water improvements, and the limitations of the current analytical
       methods for Cryptosporidium as well as other fecal indicators. However, PWSs
       that choose to employ this option, either separately or in combination with
       other approaches, may gain some benefit using this approach. For example,
       while the State and/or PWS may already have some knowledge of potential
       Cryptosporidium sources through land use information or discharge permit data,
       monitoring can help determine the extent to which these sources are impacting a
       source and can help target portions of the watershed for extra protection or BMP
       implementation. Although not required for WCP plan development,
       implementation, or maintenance, monitoring throughout the watershed for
       Cryptosporidium (or indicators of fecal contamination) can be a useful tool in
       evaluating the success of watershed control program controls WCP plan
       development, implementation, or maintenance, monitoring throughout the
       watershed for Cryptosporidium (or indicators of fecal contamination) can be
       a useful tool in evaluating the success of watershed control program controls
6. Similarly, language in Section 2.3 strongly endorses hydrologic modeling of water bodies
   and pollutant loads. Such an expectation effectively limits application of watershed
   control programs to the most sophisticated (and typically largest) drinking water systems.
7. Section should be updated to reflect revisions in implementing concentrated
   animal feeding operation (CAFO) regulations that have occurred since section was
   drafted (see
   In section, consider adding stormwater bmps as possible ways to address some
   point sources (i.e., CSOs) through reduced burden on sewer infrastructure.
8. The numbering of subsections in Section 2 should be re-worked. Numbering goes from
   2.3.3. to 2.4, then back to 2.3.4. Section, “Delineation of Area of Influence,”
   references Section 2.4.1 which does not exist.
9. Website references should be checked to make sure they are still active (e.g., Section
10. Update references that did not have a date. For example, Gullick et al., was published in
    2007 (Gullick, Richard W., Richard A. Brown, and David A. Cornwell (2007). “Source

   Water Protection for Concentrated Animal Feeding Operations: A Guide for Drinking
   Water Utilities.” AWWA Research Foundation, Denver, CO.).
11. The best management practices (BMP) section was an area were improvement is
    possible. This section is an opportunity to provide references to resources on BMPs and
    case studies of successes achieved. Appendix E and sections,, and
    overlap considerably. EPA should streamline these sections and add more references to
    Appendix E.
12. Appendix E [Watershed Control Best Management Practices (BMPs) and Case Studies]
       o First paragraph refers to section 2.4.2 which does not exist.
       o Section E.1.3, first paragraph states that “EPA’s Drinking Water State
         Revolving Fund allows a percentage of the fund to be set aside for land
         acquisition associated with watershed protection”. This should be qualified
         that some states (e.g., Pennsylvania) may not allow for funds to be used in
         this manner.
       o In Section E.2.3, consider adding stormwater BMPs as possible ways to
         address CSOs via reduced burden on sewer infrastructure.
       o In Section E.3.3., consider adding a statement about the potential for
         Urban/Suburban BMPs to reduce burden on sewage infrastructure and thus
         address some point sources (i.e., CSOs) in addition to addressing non-point
13. Appendix F [Assessment Criteria for Use By States When Reviewing Watershed Control
    Program Plan]
       o Appendix contains a checklist for States to use, but that list/Appendix is not
         mentioned in Chapter 2 (it was mentioned only in the Introduction to Section
         1.2). That checklist was Table 2.1 ("Assessment Criteria for Use By States
         When Reviewing Watershed Control Program Plans") in the original version
         of the manual, and that has now been moved to Appendix F. EPA should
         introduce the list/Appendix somewhere in Chapter 2.
       o EPA should clarify the purpose Appendix F. Appendix F includes a checklist
         of all potential tools to be used in creating a watershed control program and
         labels it "assessment criteria". Some of the actions listed in Appendix F are
         explicitly required by the rule, some are recommended guidelines in order to
         facilitate development of a WCP, and some are options which the states take
         depending on the specific situation of a PWS. Labeling these elements
         “assessment criteria” may conflict with the rule and confuse users about
         actions that are required versus actions that are simply recommended.
       o The checklist that contributes limitations to the utility of the checklist is that
         the questions are not sequenced in a logical order. The checklist does not
         proceed through a sequence of questions that lead to a resolution or
         conclusion. Completing the checklist does result in decision as to whether the
         system will be awarded the credit or not. Therefore, rather than just a
         collection of questions, the checklist needs a starting point, a clearly defined
         objective or endpoint, and a pathway between them so that utilities applying
         for the credit as well as the primacy agencies (states) reviewing the

   application will know whether the credit can be awarded or not. Equally as
   important is an understanding of the process and requirements associated with
   maintenance and continuation of the credit; in other words what the utility
   must do in order for the state to continue to award the credit.
o The Vulnerability Analysis Section needs clarification.
        i. Much of the information for larger watersheds may not be available at
           a high degree of resolution. As long information is taken into account
           in an aggregated sense then analysis remains viable and should be
           acceptable. At present the table can be misinterpreted require detail
           to a point that is unrealistic for any watershed management program
           to produce. (Row 2 in table)
       ii. The scope of the analysis must be feasible. Evaluation of activities
           within the watershed that could result in Cryptosporidium
           contamination of the water supply should be constrained to those that
           are “likely to be present and relevant and significant.” (Row 5)
      iii. Identifying the location and timing of sludge / biosolids application
           and disposal will depend on the state’s ability to track biosolids
           application and Natural Resource Conservation Service (NRCS)
           programs. This information may not be available to water suppliers
           and it is beyond the ability of most water systems to be the primary
           collector of data. (Row 9)
      iv. Locating stormwater discharges will be complicated by
          implementation schedule of CWA regulations. At present, MS4
          Phase II communities are just now putting their outfalls into GIS
          maps, consequently, in larger watershed / multi-jurisdiction
          watersheds this analysis will not be challenging to accomplish at a
          significant level of detail. (Row 10)
       v. Information on the location, age and condition of septic systems is not
          available in many areas and not required. Aggregate estimates can be
          prepared from census information estimating loadings and comparing
          sources. An aggregate analysis may find that detailed source
          information is not necessary for minor load contributors in a given
          watershed. (Row 11)
      vi. Land use is more significant than zoning and should be recognized as
          a useful analysis. (Row12)
      vii. Specific farming information at the parcel and farm level is protected
           private information that NRCS and USDA will not make available to
           water suppliers. Consequently, information is not available to support
           analysis of types of farming, feedlot locations, manure application,
           etc. Again, the question becomes what level of detailed information
           is needed for the task. (Row 13, 14)
     viii. Clancy et al. demonstrated that there is no concrete and universal
           relationship between bacteria and Cryptosporidium. Consequently,

                           evaluating bacterial levels of tributaries or areas of the reservoir is
                           unnecessary. (Row 19)
             o Potential Control Measures to Control Cryptosporidium Contamination
                       i. While assessment of economic feasibility is inherent to any decision
                          process. The questions before the state, e.g., the “assessment criteria”
                          are not an opportunity for the state to second guess the water system’s
                          political or economic choices.
                      ii. Similarly, the questions posed by EPA suggest that the agency does
                          not understand how local stakeholder partnerships work. People are
                          not contracted, there is voluntary cooperation, not documents. It is
                          more important to capture (1) what actions are being done and (2)
                          who will undertake particular actions. States do not need to make
                          decisions based people’s motives.
Bank Filtration [Chapter 4]
AWWA reviewers of the bank filtration chapter included nationally and internationally recognized
experts in environmental microbiology, hydrogeology, riverbank filtration, and drinking water
quality / treatment. The universal consensus of the review team was that this chapter requires
extensive revision prior to its adoption as formal EPA guidance. The manual should emphasize that
aquifer material with favorable properties in a setting with favorable water quality provides a reliable
and effective means of treatment. Expert universally agree that bank filtration is a treatment
technique with multiple water quality benefits and a technique that should be encouraged. This
guidance needs to be rewritten toward that end.

EPA’s rule requirements, skeptical tone, and statements in the draft guidance combine to create a
very negative environment for a drinking water utility to pursue bank filtration, despite the obvious
benefits to water quality. This posture is at direct odds with international expert opinion. For
example, in Germany, Carston Schmitt estimates that 16% of Germany’s drinking water was relied
on bank filtration and direct withdrawal has dropped to a minor fraction of Germany’s supply:

      “… approximately 16 % of the drinking water in Germany is produced from bank filtrate
      or infiltrate. Because of pollution, direct treatment of river water has dropped to 1 %.”1

Currently, EPA’s policy posture is such that water systems seeking to take advantage of bank
filtration are best served by pursuing a ground water well source through their state primacy agency
and being managed under the Ground Water Rule (GWR) or if testing results require it, under the
Surface Water Treatment Rule (SWTR) Ground Water Under Direct Influence (GWUDI) provisions.
This situation is a clear illustration of EPA policy impeding innovation and incorporation of
technological improvements into drinking water treatment that both regulators and CWSs agree are

The general view of AWWA’s reviewers was that this chapter provides ineffective and misleading
guidance with respect to the microbiological aspects of assessing bank filtration. It is contradictory

  Schmidt CK, Lange FT, Sacher F, Baus C, Brauch H-J, Assessing the fate of organic micropollutants during
riverbank filtration utilizing field studies and laboratory test systems, Geophysical Research Abstracts (2003)
Vol. 5, 08595

in many places and serves more to confuse than clarify. EPA has not clearly and consistently
reviewed and discussed the literature; rather the agency has selectively taken bits and pieces and
ignored other valid data. This document does not clearly discuss the advantages and drawbacks of
the various surrogates and in many instances clearly misleading information is presented. The
chapter needs to be rewritten for clarity and consistency and include clear information on the use of
the various surrogates that have been successfully used in these evaluations. This chapter should be
   1. Re-written by a practicing expert in the field of bank filtration using appropriate
      references to the literature and sound engineering practices, example publications
           a. NGWA 2nd Edition of Manual of Water Well Construction Practices,
           b. AWWA Water Well Standards, and
           c. Roscoe Moss Handbook of Water Well Development;
   2. Shortened to simply the information required to comply with the LT2ESWTR
      requirements. At present this is unnecessarily the longest chapter in the guidance
   3. Accurately reflect LT2ESWTR requirements;
   4. Recognize that surrogates for Cryptosporidium removal are imperfect but that a range
      of options are available and can be used in combination; and
   5. Focus on established engineering principles and delete those that are not, e.g. delete the
      excessive focus on scour.
Specific comments on the limitations of the current text and recommendations with respect to these
individual sections include:
   1. Currently, Chapter 4 implies that Cryptosporidium removal may drop from 3.5 log to 0.5
      or 1 log during flood episodes (see p 26). Experts familiar with the bank filtration
      literature were not able to find any data to substantiate this concern. In comparison,
      coliform occurrence has been observed during flood events, but have not been tied to
      stream levels. Typically coliform counts in surface streams increase substantially with
      flood events, while densities in RBF facilities are very low.
   2. Section 4.1, Introduction, immediately implies that bank filtration has limited application:
         “In optimal locations and under optimal conditions, bank filtration is suitable for
         accomplishing sufficient Cryptosporidium removal to partially meet the
         requirements of the Long Term 2 Enhanced Surface Water Treatment Rule.”
       Similarly, the Chapter does not give a full accounting of the treatment mechanism
       provided by bank filtration. An example in this section is the failure to recognize
       microbial processes:
          “Geologic units consisting primarily of fine-grained (e.g., clay-sized) materials
          will have higher removal but will be incapable of yielding economically
          significant water flow rates. In aquifers containing both sand-sized and finer
          grains, the presence of fine grains increases the possibility that pathogens will
          encounter a grain surface. This is because flow is slower and flow paths are longer
          than they would be in aquifers without such fine grains.”

3. Section 4.2, LT2ESWTR Compliance Requirements, The requirements for a bank
   filtration demonstration of performance (DOP) study include sampling both from the
   production well(s) and “monitoring wells” that are screened and located along the
   shortest flow path between the surface water source(s). But, SDWA regulations do not
   include a definition of “monitoring wells”. A monitoring locations at suitable sampling
   point(s) (like the under-the-river lateral, which is more in the flow path than would be a
   monitoring well for horizontal wells would provide better data than a specially
   constructed monitoring well and preserve the integrity of the aquifer. The guidance
   should be open to allow for such interpretations.
4. Section 4.2.1, Credits. As written the guidance is incorrect. 40 CFR 417(c)(4) requires
   systems with vertical wells to identify the distance between the “to the well screen” and
   surface water using the floodway boundary or 100 year flood elevation boundary as
   delineated on Federal Emergency Management Agency (FEMA) Flood Insurance Rate
   maps. As currently written, this important distinction is lost, and this inclusion is critical
   to the viability of bank filtration.
5. Section 4.2.1, Credits. As written the guidance is incorrect. CFR reads as follows:
         “Systems must extract a core from the aquifer and demonstrate that in at least 90
        percent of the core length, grains less than 1.0 mm in diameter constitute at least 10
        percent of the core material.”
    Guidance on page 4-4 reads as follows:
        “System must characterize the aquifer …
        The recovered core length must be at least 90 percent of the total depth to the
        projected bottom of the well screen and each sampled interval must be a
        composite of no more than 2 feet in length. …
        An aquifer is eligible for removal credit if at least 90% of the composited
        intervals contain sufficient fine-grained material as defined previously.”
    This is a clear example of EPA instituting a regulatory requirement through guidance as
    previously discussed in our cover letter.
    While the regulatory requirement is an oversimplification of aquifer characterization for
    purposes of bank filtration, the current guidance further exacerbates the situation by
    imposing a requirement for distribution of media over the entire depth of the well that
    does not exist within the regulatory requirement.
6. Section 4.3, Toolbox Selection Considerations, The draft manual reads
        “Bank filtration is best suited to systems that are located adjacent to rivers with
        reasonably good surface water quality and that plan to use bank filtration as one
        component of their treatment process.
    What is “reasonably good surface water quality”? What peer-reviewed data does EPA have
    to guide such a determination? In Europe, bank filtration is used on some of the most

        polluted waterways explicitly because they are so polluted and the beneficial effects of bank
        filtration on water quality are desired2.

    7. Section 4.3, Toolbox Selection Considerations, The draft manual reads:
            “…Wang et al (2000, 2002) documented high removal of Cryptosporidium
            surrogate organisms at production well sites in The Netherlands and in Louisville,
            Kentucky. There was very little occurrence of Cryptosporidium in river water
            at the Kentucky site and no Cryptosporidium was found in the well water at
            either site. The amount of Cryptosporidium removal at either site is
        Why is this information included in the manual? It appears that the agency is implying
        that bank filtration does not have appropriate supporting data and, in particular, that the
        agency is dismissing data from a specific group of researchers.
    8. Revise Section 4.3.1, Advantages and Disadvantages, to include additional advantages of
       bank filtration include:
            o Fewer chemicals added compared with conventional treatment.
            o Relative to conventional or membrane treatment processes, no waste stream is
              produced that requires addressing environmental issues related to
              management and disposal of these wastes.
            o Typically, lower energy consumption and reduced greenhouse gas generation.
    9. Revise discussion throughout Chapter and expand Section, Additional Treatment
       Steps, to reflect that that the aquifer, not just the hyporheic zone, contributes to removal
       (see page 4-10). A more complete description of treatment mechanism in bank filtration
       is important given concerns expressed regarding scour. For example, Section
       discourages the use of bank filtration by over-emphasizing scour:
             “Much of the removal of the contaminants and microbes discussed above occurs
            during the first few centimeters of the flow path, due to the significant filtering
            and sorptive capabilities of sediments in the riverbed. These sediments are often
            organic-rich, highly biologically active, and fine-grained. The effectiveness of
            bank filtration, however, may be temporarily threatened during high flows if
            this active layer is washed away or scoured. EPA suggests the potential for
            stream channel scour be evaluated during riverbank filtration site selection
            (section 4.4). Section 4.5 provides further discussion of scour and its
            implications for riverbank filtration system operation.”
    10. Section, Significance of deoxygenation is not provided, but by inclusion is
        implied to be quite high, especially given the agency’s verbiage. Does EPA possess data
        that deoxygenation is a frequent concern relative to the recognized benefits of bank
        filtration? Is the likelihood that deoxygenation will occur reflected in the agency’s
        criteria for “reasonably good water quality”?

  Carsten K. Schmidt, Frank Thomas Lange, Heinz-Jürgen Brauch, Wolfgang Kühn. Experiences with riverbank
filtration and infiltration in Germany. DVGW-Water Technology Center (TZW) Karlsruher Straße 84, D-76139
Karlsruhe, Germany. 2003.

        “In addition to clogging and scour, there are several disadvantages to bank
        filtration which utilities may wish to consider and balance against the advantages
        and cost savings described in section 4.3.1. One disadvantage is that an additional
        aeration step may be required during water treatment due to the possible
        depletion of oxygen as biological activity consumes oxygen during riverbank
        filtration pretreatment (Kuehn, et al., 2000). This oxygen depletion may lead to
        extremely anaerobic conditions over a portion of the flow path, which may
        sometimes result in the release of iron and manganese from the bank
        sediment into the flowing water. This process occurs due to a redox reaction
        which reduces iron and manganese to their water-soluble forms. This
        condition may necessitate the removal of these metals during subsequent
        treatment steps (Kuehn, et al., 2000; Tufenkji et al., 2002).”
11. Section 4.4.1, Coring, This section only includes descriptions of two drilling methods. At
    a minimum, it should be expanded to include discussions of rotasonic drilling (because of
    the relatively undisturbed cores which can be obtained) and cable tool drilling (possibly
    in combination with split-spoon sampling), another method which can be used for
    coring. Both of these methods will handle borehole stability issues that are mentioned in
    the text.
12. Section, The manual is not clear whether the rule is prescriptive in requiring all
    vertical wells (including existing) to be out of the 100 year floodplain or outside the
    floodway to achieve credits, or to what extent the manual is providing guidance on
    locating new vertical wells. In particular, it is not clear if both setback and demonstration
    requirements exist for those facilities that conduct a demonstration of performance
    Also, it is not clear from the guidance what constitutes sufficient separation as required
    by 40 CFR 141.715 and 141.717. A vertical well that is screened in different aquifers,
    using appropriate construction methodology (e.g., grout seal, conductor casing, etc.) such
    that the distance to the top of the filter pack from the riverbed is greater than 25 feet
    should meet the rule requirements. It should be deemed as compliant, even if the
    horizontal distance from the river boundary to the well casing is less than 25 feet.
13. Exhibit 4.6 is a very poor graphic, and it should be modified or deleted.
14. Section 4.7, Demonstration of Performance, It is important that the document provide
    flexibility in the development of a DOP given the inherent variability of sites where bank
    filtration is applied – each DOP needs to be tailored to site conditions.
    The current guidance text minimally discusses site conditions that are critical in assessing
    the effectiveness of bank filtration including soil grain metal oxide chemistry and the
    concentration and type of DOC. These issues should be discussed.
    The requirement of having monitoring wells along the shortest flow path may mean that
    wells will have to be installed in the river which is impractical from a logistical and
    regulatory perspective. Short-circuiting caused by such wells also needs to be
    In several places, the document emphasizes the negative impact of scour, however, recent
    work (Gupta, et. al. ES&T, 2009) shows that this may not be as significant as assumed.

                                             - 10 -
15. Section 4.7.3, Ground Water Travel and Residence Time, must be described with
    practical expectations. By setting expectations that cannot be met within a typical CWS’s
    fiscal constraints and rule compliance deadlines, this guidance prevents the use of bank
    filtration as an LT2ESWTR compliance tool. For example, determining the dilution with
    groundwater is a non-trivial challenge for many systems, and will be a dynamic condition
    – seasonally and yearly.
   The text seems to suggest that “ambient” groundwater is less susceptible to
   contamination by pathogens or other constituents of concern. This is not always the case
   - terrace gravel mining, agricultural operations (feed lots, dairies, etc.), septic systems,
   and other industrial activities need to be considered. This is especially true if the geology
   surrounding the alluvial aquifer is comprised of fractured bedrock.
16. Section 4.7.4, Surface and Ground Water Data Collection, the last paragraph of this
    section states that the DOP should provide data that ensures that “indicator organisms are
    not coming from sites other than the source river water”. This seems to require that
    indicator organisms exclusively originate in source river water, however, some indicator
    organisms exist in environments other than surface water such as soil or groundwater.
   The meaning and intent of the following statement is unclear: “The presence of
   alternative sources will invalidate any monitoring data obtained from the collection
   devices.” Do “alternative sources” refer to “local sources” mentioned in the prior
   sentence? If so, this leads to several questions. Does this statement refer to existing
   facilities? Does the presence of alternative sources alone invalidate any monitoring data?
   What are alternative sources and what criteria are used to identify alternative sources?
17. Section 4.7.5, Monitoring Tools, Microspheres should also be included as they can also
    be a representative surrogate in column and in-situ studies. See Metge, Harvey et. al.
    (Geomicrobiology Journal, 2007).
   The fate and transport of the Cryptosporidium surrogates listed is not clear and is not
   certain that they behave in a similar manner to oocysts. It is not just a matter of size and
   morphology; physiochemical parameters (such as surface charge buoyant density, etc.)
   can be more important.
   How would aerobic spore concentrations be used to evaluate pathogen removal
   efficiencies for situations where underflow dominates? For these situations, what is
   “ambient” groundwater and how will background spore concentrations in groundwater be
   determined relative to spores derived from surface water? Does it matter that spores
   could have originated from surface water miles upstream of bank filtration facilities?
   In the last paragraph of page 4-45, the text goes too far in speculating that aerobic spores
   are superior indicators because they are more mobile in the subsurface than total coliform
   and would yield a different result. In addition, the text infers that carboxylated
   microspheres would yield a similar result as spores. Column and in-situ studies at this
   site using microspheres show significant removal capacity of streambed sediments. If
   aerobic spores were used it might provide a differing assessment (as perhaps any other
   surrogate would), but would it be a more meaningful assessment? That is highly
   debatable because it is not certain that aerobic spores behave like oocysts in granular
   material and, if detected in extracted water, how would you know where they came from?
   A possibility is that the spores from the river are removed by the poorly sorted high iron
   sediments (in a low DOC [in terms of concentration and reactivity] environment) while

                                            - 11 -
    spores detected in extracted water could come from aerobic zones in the aquifer and
    overlying soil. This paragraph does not seem consistent with the text in the fourth
    paragraph of page 4-41 stating that no single surrogate organism is best.
18. Exhibit 4.7.1, Some of these cost estimates are from as far back as 2001, and some have
    no year listed for the estimate. All of the cost estimates should be updated to 2009
    estimates and noted as such.
19. Section, When re-drafting chapter 4, it should be re-written in everyday language
    of engineers and working hydrogeologists. For example terms like: alluvial aquifer, well-
    sorted, fine-grained sediments, etc. are common and well understood. Terms like “fluvial
    depositional processes” and “modern streams” are terms that have different means based
    on the context of the application and are not appropriate in this guidance.
20. Section 4.4.3. Correct per modification to address Section 4.2.1. This section, as written,
    further expands on regulatory text:
        “Collect relatively undisturbed continuous core samples from the surface to a
        depth at least equal to the projected bottom of the well screen for the proposed
        production well.”
        “If core recovery is insufficient, another well core must be obtained.”
        “Examine each 2 foot long composite sample of recovered core in a
        laboratory using sieve analysis to determine grain size distribution.”
        “If more than 10 percent of the sediments in each 2 foot long composite
        sample are less than 1.0 mm in diameter (very coarse sand), then the core
        interval from which it was taken is noted as containing a sufficient quantity of
        fine-grained material to provide adequate pathogen removal.”
        “To receive Cryptosporidium removal credit, at least 90 percent of the analyzed
        composited core intervals from the sampled aquifer will meet criterion
        number (4) above.”
21. Section Units of measure. Review the manual for consistent use of units
    consistent with the typical use in the field, e.g., distances are measured in feet or miles,
    volumes in gallons, concentrations in mg/L, etc.
22. Section Required separation Distance, This section states:
        “At most typical bank filtration locations, high log removal rates (e.g. 3.5 log
        removal over 13 m) may be expected with the surface water discharges that
        predominate during most of the year. During short flood periods, however, there
        may be substantially lower removal (e.g. 0.5 to 1.0 log removal over 13 m)
        due to scouring of the surface water–ground water interface, as discussed below
        in section 4.6.2.”
    After a review of the scientific literature, AWWA was not able to verify that the theory of
    diminished log removal can be justified based on available science. Data is available on
    coliform detections during flood surges, and these have not been tied to log removal (no
    data on the source, but likely very elevated), only presence in the infiltrate. The
    implication that Cryptosporidium removal in bank filtration decreases from 3.5 down to
    1.0 is conjecture.

                                              - 12 -
23. Section Locating Wells at Greater than Required Distances …, Section
    further illustrates that this Chapter should be re-written by a consultant familiar with the
    practical, rather than simply the theoretical constraints associated with bank filtration
    development (e.g., by an individual with practical experience developing bank filtration
    facilities in the United States.). This section states:
        “For example, if mapping the bedrock-alluvial interface and the water table at a
        particular site indicates that the aquifer is fairly thin, it is unlikely that infiltrating
        river water will be diluted by much ambient ground water.”
        “This may be useful at riverbank filtration sites, where water table layer and
        depth to bedrock can be used to determine aquifer thickness - an important
        parameter in determining how much dilution of bank filtrate with ambient
        groundwater is occurring.”
    Thinness isn’t as much the issue as that many of these systems lie in glacial outwash, and
    the boundary walls are fairly well defined, so there is very little lateral movement of
    groundwater towards the well because of a vertical rock valley wall.
    While thickness affects how much water you can draw into a well, the spatial variance in
    conductivity and presence of boundary walls, rainfall, and other sources of ground water
    recharge are more likely to affect ground water flow. For example, in the Florida
    panhandle, the aquifer is particularly thin at the stream, yet a high amount of recharge
    occurs from adjacent higher elevation areas of very porous soils.
    The section goes on to focus on variability in the saturated zone and linking that concern
    to a single aspect of design, distance between the source and the well:
        “When the aquifer contains fine-grained material, it is possible that well over-
        pumping may break the hydraulic connection between ground water and surface
        water, yielding a variably saturated zone underneath a perched stream, as shown
        below in Exhibit 4.5. … If possible, the potential for formation of a variably
        saturated zone can be investigated in order to provide additional information
        regarding the desirability of locating wells at greater than required distances
        from the surface water source.”
    Considerations surrounding maintaining an installation and operating regime within
    appropriately bounded saturated conditions are inherent to sound bank filtration system
    design. Rather than point to the literature for appropriate references for design, this
    manual text frames the issue as a regulatory barrier to bank filtration. As the agency is
    aware, production volume typically decreases with increased setback distance,
    consequently, using this single tool to manage the phenomena described rapidly makes
    bank filtration less economically attractive.
    Note. The referenced graphic (exhibit 4.5) is illegible.
    Section describes uses of seismic hydrogeologic investigation methods as being
    suitable for:
       “●   Estimate depth to bedrock (ideal for riverbank filtration applications).
        •   Determine the nature of bedrock (e.g., cavernous) or location of cavities. Note
            that karst buried by alluvium may contain unexpected ground water

                                                - 13 -
       •   Determine the location of faults that may juxtapose bedrock against alluvial
       •   Determine stratigraphy (useful where sands and clays may be interlayered).
       •   Determine porosity.
       •   Determine ground water particle velocities (an important parameter for
           riverbank filtration systems).”
   Neither determination of porosity or ground water particle velocities are routine
   applications of seismic methods in geotechnical engineering.

   Section describes uses of electromagnetic (EM) investigation methods as being
   suitable for understanding subsurface flows in bank filtration systems:
       “Electromagnetic (EM) methods have been used in groundwater investigations to
       delineate contaminant plumes, and thus can be useful in conceptualizing flow
       systems in a riverbank filtration context when the quality of infiltrating river
       water is especially poor. Pulse-transient EM (TEM) surveys (a type of EM
       method) may be useful in conceptualizing flow for riverbank filtration systems
       where infiltrating water quality is poor. It may also be useful in monitoring the
       quality of infiltrating water. When data is available from both borehole and
       surface instruments, EM and electrical methods can be used to map subsurface
       geology such as the locations of coarse-grained and fine-grained units.”
   This section is a mis-application of EM. If water quality is so poor that EM cannot be
   used effectively then the location is unlikely to be used as a public water supply. This
   section text should be removed, as it may lead uninformed readers to the conclusion that
   EM methods can be used to monitor subtle changes in water quality, a task to which it is
   not well suited.
24. Section, Delineating the Edge of the Surface Water Source. As noted previously,
    this guidance must be consistent with LT2ESWTR regulatory requirements. The
    requirements are:
       “For vertical wells, the ground water flow path is the measured distance from the
       edge of the surface water body under high flow conditions (determined by the
       100 year floodplain elevation boundary or by the floodway, as defined in
       Federal Emergency Management Agency flood hazard maps) to the well screen.
       For horizontal wells, the ground water flow path is the measured distance
       from the bed of the river under normal flow conditions to the closest
       horizontal well lateral screen.”
   Most bank filtration facilities are located within the 100 year floodplain on pedestals.
   Effective well construction and operating protocol can reduce the risk of contamination
   during floods. As written, the current guidance:
       o Is inappropriately ambiguous as the regulatory text is clear that the separation
         distance is measured between the surface water body and the well screen.
       o Inappropriately directive, by purposefully directing readers toward use of the
         100-year floodplain as the delineation of the surface water body, and thereby
         effectively precluding use of bank filtration in most locales.

                                           - 14 -
       o Mischaracterizes the criticality of considering the floodway to delineate the
         surface water body.
   Systems seriously considering vertical wells for bank filtration are very likely to employ
   the floodway delineation in locating wells. This delineation will have bank filtration well
   installations outside the floodway but within the 100-year floodplain. This guidance
   should direct readers to appropriate construction guides for wells build within the 100-
   year floodplain.
25. Section and, Typographical errors.
   Broken sentence:
       “The following website can be used to order these maps: flood (i.e. the 100-year flood) without increasing
       flood levels by more than 1.0 foot. It is determined by specified methods
       according to FEMA guidelines, as described below.”
   Missing verb:
       “For simplicity, if the well if closer to being a vertical.”
26. Section, Measuring Separation Distances for Horizontal Wells … . As noted
    previously, this guidance must be consistent with LT2ESWTR regulatory requirements.
    The requirement for turbidity monitoring for bank filtration reads:
           “Systems must monitor each wellhead for turbidity at least once every four
       hours while the bank filtration process is in operation. If monthly average
       turbidity levels, based on daily maximum values in the well, exceed 1 NTU, the
       system must report this result to the State and conduct an assessment within 30
       days to determine the cause of the high turbidity levels in the well. If the State
       determines that microbial removal has been compromised, the State may revoke
       treatment credit until the system implements corrective actions approved by the
       State to remediate the problem.”
   The guidance document incorrectly states:
        “To ensure that the assigned log removal credit is realized, systems are
       expected to perform continuous turbidity monitoring for all wells that
       receive a credit. Continuous turbidity monitoring is discussed in section 4.2.2.”
   Section 4.2.2 discusses turbidity measurements every 4 hours, or continuous turbidity
   monitoring, but Section inappropriately creates the regulatory expectation for all
   systems employing bank filtration for LT2ESWTR to do continuous monitoring at every
   wellhead for turbidity.
27. Like Section, Section 4.6.5 is inappropriately focused on bed scour and, in this
    instance, over emphasizes stream migration as a threat to bank filtration performance. As
    noted above regarding section, this chapter should be rewritten by an individual
    with practical experience developing bank filtration facilities in the United States.
       “Alluvial rivers that are experiencing active, progressive erosion as an adjustment
       to new flooding regimes or sediment loads, or in relation to natural lateral
       migration, may pose serious, longer-term challenges to bank filtration
       systems. For example, significant log removal reductions may be more
       frequent in an urbanizing basin as a consequence of more frequent flooding

                                              - 15 -
       and associated scouring. In extreme cases, long term degradation of the bed or
       banks may reduce the threshold separation distances between the surface water
       source and bank filtration well. Recall that these separation distances - 25 feet for
       0.5 log removal credit and 50 feet for 1.0 log removal credit - are required to
       receive log removal credits under the LT2ESWTR.”
   As a practical matter, separation distances in LT2ESWTR for bank filtration well
   installation, the nominal topic for this text, are based on floodway and 100-year
   floodplain boundaries. Both are typically over-bank limits, which are only marginally
   impacted by eroding banks. Indeed, morphology changes can make floodways narrower
   or move the channel away from the well installation.
28. Section 4.7, Demonstration of Performance, has the stated objectives:
       “… provide additional guidance on the design and conduct of a demonstration of
       performance study as well as guidance on the interpretation of the study data and
       the award of Cryptosporidium removal credits, if warranted. Finally, this chapter
       describes the necessity for long term performance evaluation monitoring to
       determine if the log removal credit continues to be appropriate.”
   As noted previously, this chapter should be rewritten by an individual with practical
   experience developing bank filtration facilities in the United States. Section 4.7 does not
   provide information that reflects:
       o A practical understanding of the costs or feasibility of the tools suggested,
       o An understanding of the regulatory requirements,
       o A clear prioritization of what is information that is important versus ancillary
         topics of marginal relevance, or
       o An understanding of the microbiology of surrogate organisms described.
   Information included without attention to practical import:
       “Environmental tracer data (isotopes, CFCs, pharmaceutical compounds,
       etc.) should be collected to verify lag times calculated using temperature,
       chloride and other parameters. …
   The mentioned environmental tracers are both expensive and of uncertain quality (e.g.,
   CFCs and pharmaceutical compounds). Such tracers have little likelihood of providing
   conclusive data where travel times are short. Perhaps more importantly, there are much
   cheaper and effective ways to collect meaningful data, using water quality parameters
   such as temperature, hardness, and fluoride. Temperature, chloride, and bromide are
   recognized but the guidance does not say that these more applicable parameters “should
   be collected.” Equally importantly, this text sends the message policy that isotopes,
   CFCs and pharmaceuticals are reasonable tracers for bank filtration facilities with
   complex flow lines. This takeaway message is wrong, and if needed modeling is a more
   reasonable method to estimate the shortest flow line than measuring for exotic
        “Late summer or drought low flow conditions should also be more intensively
       sampled if the low water levels represent a possible worst case scenario.”
   In the interest of investing samples at the most important time, a single sample in the
   drought condition would suffice, as the conditions leading up to the event are gradual and

                                            - 16 -
cumulative (e.g., flow lines steepen, velocities increase, and media becomes un-saturated
gradually over time). This is not the case in flood surge events, where hydraulics can
change dramatically over a period of hours.
    “The study design should ensure that the number and location of river water
    samples collected are representative of high and low consumptive use (e.g.
    pumping for drinking water supply, irrigation, etc.) periods. River water
    samples should be representative of the entire river volume, rather than
    consisting only of samples collected at the surface water intake for the treatment
    plant. If point sources discharge upstream and the stream is not well mixed, then
    the river samples should be proportionate in number and location to the
    volume of the highly concentrated plumes emanating from the point
This section should be redrafted to reflect the task at hand, demonstrating removal.
Language that imparts the impression that the guidance is a regulatory requirement
should be removed, and concepts inconsistent with the LT2ESWTR regulatory
requirements must be deleted. With these changes in mind, the task is best described as
sampling the concentration of the monitored parameter in the surface water supply as
near to the area of infiltration as possible and the concentration of the water drawn from
the well. A more elaborate monitoring scheme not only unnecessary to support the
LT2ESWTR framework, it is unlikely to provide significant additional value-added
information. These samples can be dangerous to get in high-flow situations…but the
point is to let the reader know what the ideal sample would be, and then to get the best
sample possible given site and safety considerations. EPA does not have the need or
authority to impose worst-case sampling or distributional sampling structures within the
guidance document.
Section 4.7.4 should focus on a practical list of water quality data. It currently reads:
    “Suitable parameters measured could include, but are not limited to, organic
    carbon, chloride, bromide, TDS, hydrogen, oxygen, uranium and other
    isotopes, and CFCs.”
A more practical list would include temperature, hardness and dissolved (oxygen) rather
than “hydrogen, oxygen, uranium, and other isotopes”. It could also point out that
chloride, hardness, temperature, and dissolved oxygen (DO) represent the central
elements of this data set. The discussion of dissolved oxygen should note the challenges
of monitoring when levels are below 3 mg/l and effects of atmospheric exposure in the
sampling well and during collection.
Section 4.7.4 also sets numerous additional regulatory requirements for DOP data collection
that are not found in the LT2ESWTR:
    “Data collection activities should be designed to ensure that the collected
    samples are representative and random. Data analyses should include
    quantitative assessment of the uncertainty associated with each conclusion.
    Study design should include sufficient sample numbers so as to determine
    statistical significance for each conclusion to a pre-determined confidence
    The study design should also include a quality assurance project plan,
    identifying 1) reference to the analytical method and laboratory, 2) a reasonable

                                          - 17 -
    number and percent of blank, replicate and spiked samples, 3) detection limits,
    and 4) sample holding times.
    The presence of multiple data collection wells can serve to increase confidence in
    the conclusions. Monitoring well data (preferably from multiple wells) collected
    along the flow path must show a decrease in indicator concentration with distance
    from surface water to improve overall confidence that the measured log removal
    results are meaningful.
    The DOP should determine the capture zone of each collection device and/or
    conduct dye trace studies from local sources such as septic tank leach fields to
    ensure that indicator organisms are not coming from sites other than the source
    river water. The presence of alternative sources will invalidate any
    monitoring data obtained from the collection devices.”
This text covers a number of useful concepts, but none are required by the rule and as
included here give the appearance that the agency is attempting to erect “roadblocks” to
the use of bank filtration.
Examples of information included of marginal relevance:
    o It is not clear what section 4.7.1 is intended to communicate. Suggest
    o Section 4.7.2, cannot be represent the agency’s views on sampling as it is at
      odds with the sampling premise underlying LT2ESWTR’s risk-based rule
      structure. As proposed, Section 4.7.2 is neither premised on representative
      loadings as LT2ESWTR is based, nor does it reflect worker safety as
      LT2ESWTR monitoring provisions clearly address. Suggest deletion.
Examples of information of uncertain technical content:
    o Section 4.7.2 refers to “stream tubes” in the context of river water quality
      sampling. What are stream tubes? Suggest deletion.
Section 4.7.5, Monitoring Tools continues to establish requirements that are not included
in LT2ESWTR regulatory text:
    “The DOP study should consist of monitoring for Cryptosporidium or a
    suite of Cryptosporidium surrogate organisms at each collection device (or
    device type cluster) and the source river water. Pathogen monitoring could
    also include Giardia and perhaps members of the Microsporidia family
    (Brusseau et al., 2005). In the absence of Cryptosporidum oocyst removal data
    (calculated using measurable oocyst concentrations in the river and in the
    collection device)”
Monitoring for Cryptosporidium is of little value since it is essentially impossible to find
it in bank filtrate. Even in surface waters, the levels are relatively low and the vast
majority of raw water samples are non-detect. If MPA is done, then using the Pall
Envirochek filter and analyzing a portion for Giardia and Cryptosporidium can be added
easily to the analysis and while it is expected that no cysts or oocysts will be detected, it
can be easily demonstrated. While pathogen monitoring could include Giardia, it is not
the same size or shape as Cryptosporidium, and finding it is as unlikely. One has to rely
on the surrogates to develop meaningful data.

                                          - 18 -
Remove the Microsporidia reference as there are no suitable references for microsporidia
methods, nor are there validated commercial kits. The paper referenced was a single
research study and few (almost no) labs are able to detect microsporidia with the
polyclonal antibody kits available. The polyclonal antibodies are not specific when
analyzing environmental samples. There are thousands of species of microsporidia and
their presence of microsporidia spores, which are approximately 1 µm in size, do not
indicate a risk of Cryptosporidium. Including this in the guidance indicates that it could
be a reliable surrogate when it is not.
There is a disconnect here with the tables that follow. Algae are the most useful indicator
when comparing surface and filtered water as they are in the highest concentration in
surface water and will always be abundant at the 5 to 7 log level. The other surrogates
may or may not be present in surface water or at a level where log reductions could be
calculated. Algae are a surrogate for surface water, whereas E. coli and fecal strep are
indicators of fecal pollution that may be transient in surface water.
The agency once more goes beyond the regulatory requirements of the LT2ESWTR in
determining what states may not use as surrogates in bank filtration DOP studies:
    “Log removal calculations for particles or organisms that significantly differ from
    oocysts in size, shape, and porous media transport capability or have unknown
    size and shape (and charge), such as turbidity, standards particle counts, and total
    algae, or larger organisms such as rotifers, crustaceans or fish are not meaningful
    and must not be used.”.
This language imposes a specific regulatory requirement where none exists in the
regulation. This directive language is also inconsistent with subsequent supporting text
and the best available science in the peer-reviewed literature.
The costs for the surrogates likely range from $50 for bacteria to $500 for MPA with
Giardia and Cryptosporidium. While EPA recommends the cheaper assays, there is no
evidence correlating total coliform, aerobic spores, and enterococci with transport and
removal of Cryptosporidium in bank filtration. To effectively determine log removal of
a biological surrogate, the most important criterion is that it be present consistently and
at high concentrations in the source water. If the bank filtrate samples are non-detects,
it is with some assurance that the indicator was truly not there as opposed to absent
because it was at the limit of detection. For example, if a surrogate is present at 1 to 2
log in the source, and not detected in the bank filtrate, this could be due to sampling or
working at the limit of detection of the test method. If a surrogate is present in the 5 to
7 log concentration in the source, detection if truly present in the bank filtrate is more
reliable; a non-detect is more reliable in these circumstances.

The cost table, Exhibit 4.7.1, should be removed. Costs can vary more than 3-4 times
from lab to lab; the method cost itself is meaningless as no laboratory can set up any
TC method for 71 cents. Data from 2001 and 2005 are poor indicators of 2009 costs.
The table adds nothing and could confuse those trying to procure services at those

Section 4.7.5 and much of the bank filtration chapter follows from the (1) the agency’s
views on in situ Cryptosporidium and surrogate monitoring data and (2) the
microbiological data to determine if and how much removal is occurring at a particular

                                         - 19 -
bank filtration installation. In drafting the LT2ESWTR, EPA did not reserve the right
to dictate which microbiological surrogates could or should not be used. With this
absence of authority in mind, section 4.7.5 should be rewritten:
   o Exhibit 4.7.2 is incorrect and misleading. It should be removed.
   o Criteria describing surrogate indicators for Cryptosporidium should be
     removed, e.g. what ‘slightly oblate’ means in describing the shape of oocysts
     for the purposes in bank filtration is not very useful. There are algae that are
     very similar in size and shape, but the bacterial indicators suggested do not
     meet these criteria.
   o An emphasis should be put on identifying an indicator that is present in
     sufficient quantity and can be sampled in a large enough volume, before and
     after bank filtration, as to allow calculation of log removal.
   o There are several available but imperfect surrogates, as the available data
     illustrate that in different situations, it may take one or more types of data to
     present a sufficiently robust analysis; consequently, EPA guidance should be
     equally open to a spectrum of surrogates.
Making the above changes are necessary for the guidance to have any credibility. At
present, a brief review of basic references (e.g., Bergey's Manual of Determinative
Bacteriology (Holt,1986)) find that the surrogates identified in the guidance do not meet
the criteria the agency is articulating:
   o Family Enterobacteriaceae – are straight rods, occurring singly, in pairs or in
     short chains; they are usually 0.3 to 1.8µm in size. The five most common
     genera isolated in water are described as follows: Citrobacter – 1 x 2-6 µm,
     Enterobacter – 0.6 -1.0 x 1.2-3.0 µm, Escherichia – 1.1-1.5 x 2.0-6.0 µm,
     Klebsiella – 0.3-1.0 x 0.6-6.0 µm, and Serratia – 0.5-0.8 x 0.9-2.0 µm.
   o Coliform bacteria are rod-shaped any may produce extracellular capsular
     material; they are not like oocysts in size or shape since oocysts are basically
     round. A defining characteristic of the coliform bacteria is that they are rod-
     shaped. While in some cases they may be as long as an oocyst is wide, they
     are always rods.
   o Giardia cysts are in no way similar to E. coli in size and shape.
   o C. perfringens are in no way similar to Cyclospora and
   o Clostridial spores are in no way similar to Microsporidia.
The agency’s treatment of spores in Section 4.7.5 is controversial and does not represent
consensus expert judgment or a robust and balanced summary of the peer-reviewed
   o Aerobic and anaerobic spores are present in low numbers in surface and
     ground water. An overemphasis on spores as indicators runs the risk of
     forcing the use of a single tool that may not be able to demonstrate
     meaningful levels of removal at a particular site.
   o Aerobic and anaerobic spores are known to be extremely long-lived in the
     environment, consequently it is not clear when they are used in bank filtration
     assessments if the spores are from the surface water source or ground water

                                        - 20 -
     with a much longer travel time. At low concentrations, even a small
     concentration of additional spores can substantially alter recognized log
   o There is an important distinction between the use of spores in pilot plant and
     full-scale demonstration of a conventional treatment plant and bank filtration
     – once the water containing the initial concentration of spores enters the
     concrete confines of the water treatment plant, no additional spores will enter
     the treated water during treatment; this is not true of bank filtration.
   o A high detection limit can be avoided in the spore assay by filtering the
     100mL sample.
EPA’s treatment of the literature must be fair, balanced, and transparent:
   o It is incorrect to state that “aerobic spores have long been recognized as a
     useful measure of surface water influence on and hygienic quality of ground
     water” citing the single reference Schubert (1975). The nature of spore
     survival in the environment, their common occurrence in soils and sediments,
     and their lack of correlation with fecal contamination do not indicate anything
     about the hygienic quality of ground water. The literature in support of this is
   o EPA states that “the aerobic spore natural background concentration is 10
     CFU/100 ml or less. Values higher than 100 CFU/100 ml may be considered
     to have some surface water influence.” There is no scientific basis for this
   o EPA needs to distinguish between conventional treatment plant performance
     and bank filtration in its literature citations.
   o EPA should not mis-represent the findings of peer-reviewed literature (e.g.,
     Locas et al (2008), Mazoua and Chauveheid (2005), Rice et al (1999), and
     Schubert (1975)).
   o On page 4-43, EPA cites unpublished EPA work evaluating laboratory
     performance with aerobic spores. This study should be made available to the
     public. Now that the report is in the public domain by reference in this draft
     guidance, AWWA requests that agency staff forward a copy of the study
     design, study data, analysis, and reports to AWWA’s Government Affairs
     Office, to the attention of Steve Via.

There are several useful ways to employ algae in this analysis and the current guidance is
inconsistent and neglects available tools:
   o This chapter should rely on all peer-reviewed materials, not select ones.
   o The Reilly et al (2005) study used the presence of diatoms, all of which are
     much larger than Cryptosporidium oocysts (4-6 µm), to indicate surface water
     influence on wells. If diatoms are useful, then as a subclass of algae, other
     algae are also useful in measuring the effectiveness of bank filtration.
   o In acknowledging that no single Cryptosporidium surrogate is best, EPA cites
     the Reilly et al (2005) reference for using diatoms as surrogates for surface
     water influence of ground water. Also cited should be the Gollnitz et al
     references, cited elsewhere in the chapter, where algae are used for this

                                         - 21 -
              purpose. An additional reference - Gollnitz, W.D. et al. 1997. A proposed
              method for evaluating natural reduction of microscopic particulates in
              alluvial-valley aquifers. J. Am. Water Works Assoc. 89 (11):84-93 should be
              added as it discusses using diatoms and algae to assess removal in bank
          o There is no valid scientific reason to exclude total algae used by Gollnitz et al
            (1994, 1997, 1999, 2005, 2007) as stated on page 4-41.
          o While EPA states that it is unacceptable to use total algae for examining the
            bank filtration process, MPA is included as one of the surrogates; algae are a
            major parameter considered when using MPA.
          o Available data do not indicate that diatoms are more useful than algae as
            surrogates for Cryptosporidium. Diatoms are found consistently and at high
            concentrations (104 to 107 per 100L) in surface waters; oocysts are found
            rarely and never at levels reaching those of diatoms. Diatoms are much larger
            than oocysts and finding a single diatom in a ground water sample does not
            mean that it is likely that oocysts could be present.
       Examples of information of uncertain technical content:
          o The suggestion that “After several months residence time in the subsurface, it
            is likely that their green color will fade” is not supported by any reference.
            Diatoms are found in ground water collection devices and there is no
            indication as to how long they have been there.
          o The Walker et al (2005) reference should be removed since the sentence goes
            on to indicate that immunoassay is not useful for this purpose due to

Pre-Sedimentation [Chapter 5]

Specific comments and recommendations for Chapter 5 include:
   1. Section 141.717(3)(i) is the proper citation for Section 5.2.3. of Chapter 5.
          “The presedimentation basin must achieve the performance criteria in paragraph
          (3)(i) or (ii) of this section.
              (i) Demonstrates at least 0.5-log mean reduction of influent turbidity. This reduction
          must be determined using daily turbidity measurements in the presedimentation process
          influent and effluent and must be calculated as follows: log10(monthly mean of daily
          influent turbidity)--log10(monthly mean of daily effluent turbidity).
              (ii) Complies with State-approved performance criteria that demonstrate at least
          0.5-log mean removal of micron-sized particulate material through the
          presedimentation process.”
       Turbidity is ill suited to serve as an indicator of presedimentation basin performance at many
       facilities, consequently the state-approved performance criteria described in Section
       141.717(3)(ii) is particularly important. Unfortunately it is unaddressed in the current
       guidance. Appendix 1 specifically addresses demonstration of performance but it contains
       several concepts of importance to this chapter as well:

                                                - 22 -
           o Overcoming the frailties of incomplete datasets can be accomplished through
             alternative central tendency estimates and appropriate data rules.
           o Use of spores rather than turbidity to estimate removal.
   2. Table 5.1 requires careful reading of text to understand, and can be taken out of context.
      Limit to text description or clearly describe table contents in titling, headings, and footnotes.
   3. 5.3.2, Advantages and Disadvantages of Installing a Presedimentation Basin, the last
      sentence should be deleted; it is not supported by data and is often not true.
           “The presedimentation process can reduce influent fluctuations in particle loading,
           flow, and other water quality parameters. An additional sedimentation process in
           series provides increased operational flexibility to handle rapid changes in influent
           turbidity. It also allows for enhanced performance of subsequent processes in the
           treatment plant. Although, if the presedimentation effluent turbidity is too low,
           the second sedimentation process may not be able to provide significant
           removal since removal performance is enhanced by increased particle load.”
Lime Softening [Chapter 6]
Specific comments and recommendations for Chapter 6 include:
   1. Section 6.2.1: Graphic is erroneous. Coagulant is not required. Soda ash is not often used or
      required in the second stage. Flocculation is not required and should not be implied.
      Recarbonation is not required and should not be implied.

   2. Section 6.2.2: Guidance on softening is premised on a misunderstanding of the softening
      process and the LT2ESWTR requirements for softening. Coagulant is not required and EPA
      acknowledges in the rule that it is not necessary.
           “In addition, EPA recommends submitting a schematic of the treatment process
           to the State, clearly identifying the two stages of clarification. EPA also
           recommends that systems monitor the coagulant dosages (or concentration)
           in the secondary clarifier on a daily basis, for the first year, and record the
           average and minimum coagulant concentrations. This data can assist the
           State in assessing whether the system operates in compliance at all times.”
           Actual LT2ESWTR language regarding softening reads as follows:
           “Two-stage lime softening is a process in which chemical addition and
           hardness precipitation occur in each of two distinct unit clarification processes
           in series prior to filtration.” 40 CFR 141.2.
           “0.5-log credit for two-stage softening where chemical addition and hardness
           precipitation occur in both stages. …” 40 CFR 141.715
               “(b) Two-stage lime softening. Systems receive an additional 0.5-log
           Cryptosporidium treatment credit for a two-stage lime softening plant if
           chemical addition and hardness precipitation occur in two separate and
           sequential softening stages prior to filtration. Both softening stages must treat
           the entire plant flow taken from a surface water or GWUDI source.” 40 CFR

                                                - 23 -
Combined and Individual Filter Performance [Chapter 7]

Specific comments and recommendations for Chapter 4 include:
   1. Section 7.2.1 of the document begins to list the EPA regulation reference as 40 CFR
      141.727. At US EPA’s website, the reference location for the regulation is 40 CFR
      141.718 Treatment Performance Toolbox Components. There is no regulation beyond
      141.723. Perhaps this reference is from a prior draft regulation. Please verify and correct
      as required throughout guidance documents.
   2. In approximately 20 pages of guidance on combined and individual filter performance, the
      agency has included more than 40 actions a water system “should” do in addition to actually
      complying with the regulatory requirement as required to comply with the LT2ESWTR
      provisions. The agency should review this chapter and the entire guidance manual to
      eliminate instructions that expand the actual or perceived requirements described in
   3. The guidance must be consistent with the LT2ESWTR regulatory requirements. In Section
      7.2.1, the guidance states:
          “1) IFE turbidity must be less than 0.1 NTU in at least 95 % of the maximum
          daily values……..”
      The regulation states that the
          “IFE turbidity must be less than 0.15 NTU in at least 95 % of the maximum daily
          values …”
      This error is also stated in section and 7.3.2 with respect to individual filter effluent.
   4. Exhibits 7.1 and 7.2, the tables on maintenance and calibration of turbidimeters should be
      deleted. The text indicates that systems should follow manufacturer’s procedures, but then
      provides these tables. The tables are rather confusing and the states can easily make the
      mistake of using these as a “shall,” rather than appropriately following instrument specific
      manufacturer’s procedures.
          o In exhibit 7.1, on-line turbidimeter cleaning, flow verification, and secondary
            standard calibration verification are “recommended” weekly. This
            recommendation is not consistent with the ESWTR guidance (e.g. monthly).
            There was an extensive data collection effort through ASTM to support on-line
            method development which also informed the ESWTR guidance development.
            There has no been any subsequent data collection on which to base an increased
            frequency or demonstrate benefits from increasing cleaning and secondary
            standard verification.
          o AWWA’s member’s experience is that on-line turbidimeters do not drift within
            a week or monthly timeframe. A monthly cleaning and standard check is
            satisfactory in achieving consistent performance from an on-line turbidimeter.
            The monthly standard check is rarely out of acceptable range. Increasing the
            frequency of the cleanings and calibration verification to weekly will not
            improve the reliability of the filter performance and will not provide any
            increase to public health protection.

                                                 - 24 -
           o For filters operating at the turbidity levels specified to meet these filter
             performance credits, the water is so clean that increased cleaning frequency is
             counter to the logic of the typical maintenance requirements for these
             instruments. To meet the 0.15 NTU requirements for individual filters, these
             filters must operate at or below 0.10 NTU to insure the factor of safety required
             to meet the regulations. Process operating tools must be optimized for every
             situation and operator response capability must insure filter shutdown within 15
             minutes of an alarm condition. The only logical reason for increasing the
             maintenance frequency to weekly is that light sources are expected to decay in
             strength over time. However, experience indicates that the impact of light
             source decay does not occur within a monthly calibration check time frame.
       Using a single large metropolitan water system with two plants as an example, the increased
       maintenance associated with the proposed guidance manual would result in the hiring of
       three additional technicians in to provide the weekly maintenance on 168 online
       turbidimeters. The total annual cost would be $210,000 with benefits. This increased cost
       would provide no benefit to the public and would not improve the system’s ability to comply
       with this toolbox credit’s performance criteria. If EPA pursues increasing on-line
       turbidimeter maintenance from monthly to weekly, it must provide sound justification for this
   5. Section, Filter Beds. Underdrain examination, as the term is typically understood in
      the water utility community, is extremely difficult to do as it involves taking the filter out of
      service, removing the filter media, and examining the underdrain without causing extensive
      damage to the underdrain. Similarly, use of the term “regular” typically calls to mind a
      routine frequency measured in days or months. Consequently, the recommendation that
      “Underdrains should also be examined regularly” is at odds with practice at even the best
      operating drinking water treatment plants. It seems likely that the authors have
      misinterpreted existing guidance for evaluating filter media. This item should be deleted or
      reframed in terms of filter media examination.

Second Stage Filtration [Second Stage Filtration]

   1. Section 9.2.1, Credits is incorrect. The regulatory requirement for second stage filtration
      receiving removal credit is:
           “Second stage filtration. Systems receive 0.5-log Cryptosporidium treatment
           credit for a separate second stage of filtration that consists of sand, dual media,
           GAC, or other fine grain media following granular media filtration if the State
           approves. To be eligible for this credit, the first stage of filtration must be
           preceded by a coagulation step and both filtration stages must treat the entire
           plant flow taken from a surface water or GWUDI source. A cap, such as GAC, on
           a single stage of filtration is not eligible for this credit. The State must approve
           the treatment credit based on an assessment of the design characteristics of
           the filtration process.”
   2. Section 9.4.3, Turbidity Monitoring is at odds with LT2ESWTR’s regulatory
      requirements. EPA should limit the guidance to requirements of the rule and not expand
      in this instance to operational criteria that were not included in the regulatory

                                                 - 25 -
           Guidance reads:
           “EPA recommends monitoring the turbidity of the individual filters in the
           second stage in order to be able to identify any possible filter upset situations.
           Depending on the first filtration stage effluent quality, it may be difficult to
           see a significant difference in the second stage effluent. If the combined
           second stage filter effluent is the only process stream monitored, it is unlikely
           that an upset in one second stage filter could be detected.”

Ozone [Chapter 11]

Comments on the ozone chapter and associated appendices will follow in a separate submittal.

Demonstration of Performance [Chapter 12]

In commenting on the proposed LT2ESWTR in January 2004, AWWA submitted comments on the
Demonstration of Performance (DOP) provisions of the regulation and associated guidance. Those
comments were prepared by Environmental Engineering and Technologies, Inc.. That document like
the draft guidance manual is supportive of aerobic spores as a surrogate for Cryptosporidium oocysts.
It also addressed a number of issues that are not addressed by the draft guidance manual and
documented references. A copy of this document is provided as an attachment here for your use.
Specific comments and recommendations for Chapter 12 include:
   1. Additional references that warrant consideration and inclusion in the premise for the :
           o Brown, Richard A.; Cornwell, David A., Using Spore Removal to Monitor
             Plant Performance for Cryptosporidium Removal, Journal American Water
             Works Association, Vol. 99, Issue 3, March 2007, Pages 95-109.
           o Cornwell, David A.; MacPhee, Michael J.;Brown, Richard A.; Via, Steve H.,
             Demonstrating Cryptosporidium Removal Using Spore Monitoring at Lime-
             Softening Plants, Journal American Water Works Association, Vol. 95, Issue
             5, May 2003, Pages 124-133.
   2. Because the DOP involves aerobic spores extensively, a reference to the Standard
      Method for spores that Eugene Rice from USEPA prepared (Standard Method 9218,
      2001) is warranted.
   3. Calculation procedures, like the rest of the chapter, are out of date (see Equation 1 and
      Figure 4 of March 2007 spore paper)
   4. The draft guidance manual does not correctly frame DOP guidance. As written, EPA
      treats DOP as an “add-on” credit, not an “adjustment” credit. Therefore, a water
      treatment plant has to prove that it is adding something extra to the plant, and to keep
      proving the extra treatment added is still there. This is at odds with the theory behind
      DOP. DOP is not about adding anything new to the existing plant rather what the system
      is doing is establishing that the current facility, without any improvements, is actually
      achieving more than baseline 3 credit assumption underlying LT2ESWTR’s treatment

                                                - 26 -
       Therefore, modification to the DOP chapter is needed to appropriately portray the credit
       as an adjustment to the baseline credit, not an additional component added onto the
       baseline credit.
           o Literature data shows that when a utility produces performance equivalent to
             IESWTR compliance (CFE <0.3 ntu >95% of the time) the facility can
             remove 4, 5, or more log of Cryptosporidium (see Table 1 of Brown and
             Cornwell, March 2007 JAWWA).
           o Even though many water treatment plants may have better performance, the
             data justified USEPA assigning a 3.0 baseline credit for complying with
             IESWTR even if the system did not do anything more to prove that level of
             treatment beyond maintaining IESWTR compliance.
           o There are two kinds of credits in the Toolbox – most are “add on” meaning
             the water system adds something on to the WTP. The other credits are
             “adjustment” credits. Chief among the latter is DOP. The DOP is not
             something new or something added onto an existing facility. What DOP does
             is prove via a demonstration study that the baseline 3.0 credit is not correct
             for the particular facility involved in the DOP study – it proves that the
             facility as it currently operates, without any improvements and without
             anything added on achieves 3.0 credits when it meets the same requirements
             as needed for IESWTR compliance -- in other words, IESWTR compliance =
             whatever DOP demonstrates.
           o If DOP proves, IESWTR compliance at this facility is not worth 3.0 credits, it
             is worth 4.5 credits total, or 1.5 above the baseline credit (i.e., enough for bin
             2, plus enough for Bin 3 and 4 if you add at least 1.0 credits from UV or some
             other “upper bin technology”).

Ultraviolet Light (UV) [Chapter 13]

At a policy-level, the issue of power quality is adequately addressed in the UVDGM and needs
minimal focus in Chapter 13.
There are a number of detailed comments necessary to accurately communicate the content of this
   1. Exhibit 13.1: need to correct the UV dose table shown;
   2. Page 13-2: third two paragraphs should be bulleted like in the UVDGM;
   3. Page 13-3: Consider deleting the last bullet on the page discussion power quality because
      most operating facilities are not having issues with power quality even without power
   4. Page 13-4 line 5: replace “monitor” with “estimate”;
   5. Page 13-4 line 10: replace “dose” with “light”;
   6. Page 13-4 line 12: add “and have higher UV light intensity output” to the last sentence;
   7. Page 13-4 line 18: replace “to” with “to be”;
   8. Page 13-4 line 19: delete the word “directly”;

                                                - 27 -
   9. Page 13-4 power quality bullet: considering deleting this discussion because power
      quality is not a major design factor, and the reader can go to UVDGM to learn about this
      minor issue;
   10. Page 13-4 Hydraulics bullet: add “for a reactor” at the end of the last sentence; and
   11. Page 13-4: They could add a bullet on monitoring to be consistent with the other
       chapters that summarizes the LT2 required monitoring of UV intensity, flow, UV dose,
       UVT, off-spec water, and calibration of UV sensors.

Membrane Filtration [Chapter 14]

Specific comments and recommendations for Chapter 14 include:
   1. Section 14.3 – some UF membranes can also remove viruses (see Title 22 of California
      DHS), and it also lowers DBPs by removing precursors.
   2. Section 14.4 – last two paragraphs on page 14-4 discuss backwashing. It should be noted
      that spiral wound NF and RO backwashing is not applicable.

                                                - 28 -
                                 Attachment 1

AWWA Comments on LT2ESWTR Demonstration of Performance (DOP) Provisions and
                               Associated Guidance
(Extract from AWWA Comments on Proposed Rule, Appendix 1, Chapter 4, Prepared by
            Environmental Engineering and Technology, Inc.) January 2004
                                         CHAPTER 4


       Experimental data shows that it is possible for existing surface water treatment facilities
to achieve 4.0-log or greater Cryptosporidium removal, as discussed in Chapter 3. Although this
does not necessarily mean that the baseline automatic credit should be increased across the board
from 3.0 to 4.0 for all US facilities, it does mean that it should be recognized that many existing
US facilities do already possess Cryptosporidium protection capabilities equivalent to 4.0 or
greater total credits, and enough realistically and appropriately defined credits need to be
available to establish the greater capability demonstrated for these existing systems. Stated
another way, since IESWTR compliance for some facilities results in treatment capabilities of
4.0 credits or greater, not just 3.0, some of the toolbox credits should be thought of not as
something added to an existing facility, but rather as a correction or adjustment to the 3.0
baseline credit.

       IFE, CFE, two-stage filtration, and two-stage clarification credits are examples of toolbox
credits included to adjust the total credit for an existing facility upward from the 3.0 baseline
automatic credit (2.5 for direct filtration) to its appropriate higher level. However, the most
direct, useful, and cost effective way to establish the true Cryptosporidium removal capability of
existing systems is to demonstrate this capability in pilot-scale challenge studies or full-scale
evaluation studies included in the demonstration of performance (DOP) credit. The DOP credit is
expected to be especially advantageous for facilities in bin 2 that do not have existing facilities,
or are not planning to build new facilities, that can achieve the 4-log total requirement using
automatic credits. The DOP study can be conducted on an entire treatment process, or on a
specific segment of the process. It can include monitoring of ambient aerobic spores in full-scale
treatment processes, or pilot-scale microbial challenge (spiking) studies using Cryptosporidium,
aerobic spores, or pilot studies using a suitable microbial surrogate, or some combination of all
of these alternatives.

March 31, 2009                                                                      Page 47 of 185
        Full-scale studies using ambient aerobic spores are a preferable demonstration study
because they evaluate full-scale facility itself, plus these full-scale spore monitoring studies are
significantly less costly than pilot-scale DOP studies. However, it may not be possible to
demonstrate the full potential of the full-scale process if the raw water spore levels are not high
enough. Consequently, pilot-scale microbial challenge studies may be necessary to demonstrate
the full treatment potential of full-scale processes. Pilot-scale microbial challenge studies could
be used in at least two different ways. The most direct use of pilot-scale testing is to replace full-
scale spore monitoring studies with pilot-scale microbial challenge studies as a means of
establishing the DOP credit. However, pilot-scale studies could also be combined with full-scale
spore monitoring studies in a hybrid approach where the ratio of spore and Cryptosporidium
removal could be established in pilot-scale studies and then this ratio used as a correction factor
in full-scale spore monitoring studies. Even though these pilot-scale studies will be more
expensive than full-scale aerobic spore monitoring studies ($50-60 K versus $0.5-0.6 M), these
costs are still significantly less than some of the other toolbox alternatives, which can cost
several million dollars. Furthermore, since the costs for both full-scale and pilot-scale DOP
studies are not dependent upon size of the full-scale facility, the DOP credit becomes more cost
effective as the facility size increases.

Full-Scale Studies

        One year of weekly full-scale aerobic spore monitoring data is a suitable and appropriate
process for establishing the DOP credit. This approach is fairly straightforward, simple, and cost-
effective, especially for facilities in bin 2. Establishing this credit can also be useful for bin 3 or
4, in conjunction with an upper bin technology (like UV), where DOP provides all of the credit
needed for these bins (4 credits in bin 3 and 4.5 credits in bin 4) except for the 1.0 credit
provided by one or more upper bin technologies. The following discussion illustrates some of the
key issues with respect to using full-scale spore monitoring to establish the DOP credit,
including the following:

            •   Aerobic spore removal is a conservative indicator of Cryptosporidium removal

March 31, 2009                                                                         Page 48 of 185
         •   Approved analytical methods are available for aerobic spores that use equipment,
             materials, and procedures that are familiar and available to drinking water
             microbial laboratories

         •   Finished water detection limits for spores can be lowered by using larger sample
             volumes in order to increase the ability to mathematically demonstrate higher
             credits [higher detection limits (lower sample volumes) can underestimate true
             treatment capability of system]

         •   Use of mean to calculate the numerical value of the DOP credit will either result
             in gross miscalculation of the true spore removal (can be over- or under-
             estimated) or will require complicated statistical methods to identify and correct
             data outliers and account for censored (below DL) data. Use of median spore
             concentration data avoid these problems, and will result in a suitable and simpler
             approach to establishing the numerical value of the DOP credit.

         •   Geometric mean results in similar numerical values of the DOP credit as when
             median is used. However, geometric mean may create some difficulties or
             confusion due to numerical limitations of the geometric mean algorithms in some
             spreadsheet software.

         •   Low raw water occurrence can inhibit ability to mathematically demonstrate true
             treatment capability of systems being evaluated, but consequences of this have
             been overstated in the draft version of the LT2ESWTR and associated guidance

         •   Facilities using at least one year of weekly full-scale spore monitoring data can
             mathematically demonstrate 4.0 log or greater spore removal, and hence are
             capable of achieving at least the same level or higher of Cryptosporidium removal

         •   Spore sample collection more frequently than once per week is not necessary to
             establish the DOP credit (though facilities may voluntarily choose to collect
             samples more often)

March 31, 2009                                                                  Page 49 of 185
            •   Once established, the DOP credit should be retained as long as the facility
                maintains compliance with the IESWTR

        Some of the results listed in the following discussion are provided from available
literature sources which are cited below. Other information is provided from a pending literature
article by Cornwell and Brown summarizing a study jointly funded by AWWA and six US
drinking water utilities. Spore monitoring data was collected at six facilities described in this
document as facilities “A” through “F”. Facility “A” includes a conventional treatment facility
with two-stage filtration using a river as source water. Facilities “B” through “D” utilize two-
stage lime softening preceded by pre-sedimentation to treat a river source. Facilities “C” and “D”
incorporate polymer and ferric sulfate throughout each of their three clarification stages plus
polymer added to filter influent, but facility “B” only uses filter aide polymer to supplement the
lime softening process, though they do recycle lime softening solids from the middle clarification
stage to the pre-sedimentation basin influent. Facilities “E” and “F” use high quality lake sources
as source water, with facility “E” using conventional clarification and filtration treatment and
facility “F” using direct filtration.

Aerobic spores and other Cryptosporidium surrogates
        As illustrated in Figure 4.1, data reported in the literature for pilot-scale microbial spiking
studies indicate that Cryptosporidium is typically removed more readily than aerobic spores
during physical treatment process like clarification and filtration. Consequently, facilities that
demonstrate around 4-log removal of aerobic spores during treatment are probably capable of
greater than 4-log Cryptosporidium removal. These findings are the basis for the inclusion of
aerobic spore monitoring at full-scale facilities as an option for the demonstration of
performance (DOP) credit in the upcoming LT2ESWTR. This is also the reason why some
participating utilities in this project have already initiated spore monitoring as a tool to help them
evaluate treatment plant performance. The guidance manual should clearly state that spores are
an accepted conservative estimator or Cryptosporidium removal.

        Figure 4.1 includes data published in the literature from pilot-scale Cryptosporidium
spiking studies conducted under stable hydraulic and coagulation conditions. These results are
representative of DOP sampling which will be conducted as outlined in the LT2ESWTR to

March 31, 2009                                                                         Page 50 of 185
measure spore removal from raw water through combined filter effluent at a surface water
treatment facility operated in compliance with the IESWTR under normal conditions. Other
conditions were tested in some studies (Huck et al. 2002 and Dugan et al. 2001) to simulate
conditions of non-stable coagulation or stressing of an individual filter. Interpretation of the
“stable” and “non-stable” data reported by Huck et al. (2002) and Yates et al. (1998) is
complicated by the fact that added Cryptosporidium and spores were coagulated in a jar, then
spiked into the filter influent, which contained ambient spores coagulated in the rapid mix that
passed through the sedimentation basin. By contrast, other studies listed were conducted by
spiking Cryptosporidium into the raw water or rapid mix and using ambient or spiked spores also
coagulated in the rapid mix.

       The first preference when monitoring treatment effectiveness for a target organism would
be to monitor the target organism itself. Cryptosporidium monitoring has obvious limitations as a
tool for monitoring treatment performance because they are not always present in raw water, are
typically found at low concentrations when they are present, and analytical methods are poor. By
contrast, aerobic spores are ubiquitous in almost all source waters (especially surface water) at
all times of the year, whether Cryptosporidium is present or not, and are present in larger
concentrations than Cryptosporidium. In addition, analytical methods for enumeration of spores
are more reliable and simpler than Cryptosporidium methods, as will be discussed later. If no
Cryptosporidium are present in the raw water, Cryptosporidium monitoring does not allow
determination of treatment efficiency. Conversely, aerobic spore monitoring can evaluate the
capability for Cryptosporidium removal throughout the year under a variety of environmental
conditions even when no Cryptosporidium is present in the raw water.

       Even though monitoring of ambient aerobic spores in full-scale treatment processes is
superior to Cryptosporidium monitoring as a tool for evaluation of treatment performance, the
raw water occurrence and filtered water concentrations of aerobic spores may still cause some
difficulties in mathematically demonstrating the full capability of a treatment process for spore
removal and by inference Cryptosporidium removal. For example, even though spores are
present at higher concentrations than Cryptosporidium, the raw water spore levels may not be
high enough and the filtered water detection levels may not be low enough to mathematically
demonstrate the full potential for the treatment process to remove spores. In this case, the only

March 31, 2009                                                                    Page 51 of 185
way to demonstrate the full potential of the treatment process is to artificially increase the raw
water spore levels. Since spiking of full-scale facilities is logistically and economically
infeasible, the only way to realistically perform spiking studies is to use pilot-scale studies.

       A recent case study completed for the City of Richmond (VA) as part of an AwwaRF
tailored collaboration project included contingency plans the facility can use to comply with the
upcoming LT2ESWTR. The findings from this study (Cornwell et al. 2003) illustrate that the
DOP credit using aerobic spore monitoring is the most cost-effective alternative for the facility if
it ends up in bin 2 (~$60,000 versus several million dollars for many other alternatives, even if
they could fit these alternatives at the existing space-limited site), and that DOP would also be
useful for bins 3 or 4 as a low cost safety factor.

       Aerobic spore samples reported in this chapter were collected and analyzed using a
protocol based upon the pending Method 9218 scheduled for publication in the next version of
Standard Methods (APHA N.d.). This method is derived from a procedure originally developed
to count aerobic spores in milk products (APHA 1993), and has been previously reported by Rice
et al. (1996). This method does not limit sample size. The method explicitly allows use of 200,
500, and 1,000 mL containers during the heating step, but a sample can be distributed into
multiple identically sized containers during this step (i.e., a 5-L sample can be heated in five 1-L
sample containers).

       One of the attractive features of aerobic spore monitoring is that it involves a fairly
simple analytical method which essentially includes the membrane filtration procedure already
used by many utilities for coliform analysis, preceded by a “heating” step. The heating step kills
bacteria in the vegetative state, leaving only heat-resistant bacterial spores to be counted in the
remainder of the procedure. Many utilities already possess the equipment and expertise to
perform the filtration procedure, including the membrane filtration apparatus and an appropriate
incubator. The only new equipment needed for most facilities that do not already analyze spores
is an oscillating water bath or a stirring hotplate needed for the heating step. The only other
materials needed, which some utilities may already have, are autoclavable glass or plastic
containers suitable for the heating step. In order to get spore counts on the plates into the
countable range (20-80 spores per plate), some raw water samples and other samples early in the

March 31, 2009                                                                        Page 52 of 185
treatment process need to be diluted prior to the heating step. Container sizes used for the heating
step need to be the same for all samples and blanks. For example, this may mean a single 500
mL sterilized container will be used for a diluted raw water sample, two or more sterilized 500
mL containers for an undiluted finished water sample, and a single 500 mL container for the
temperature control blank during the heating step.

        Bacillus subtilis is a specific aerobic spore forming bacteria counted when total aerobic
spore spores are measured. Therefore, a possible alternative to counting total aerobic spores is to
only count B. subtilis. Unfortunately, since B. subtilis is a subset of total aerobic spores, counting
B. subtilis would exacerbate the mathematical difficulties associated with demonstrating target
log removals when raw water spore levels are low.

        Total anaerobic spores are another surrogate that could be used to demonstrate
Cryptosporidium removal. Since anaerobic spores behave similarly to aerobic spores during
drinking water treatment; therefore, it is unclear what additional benefit can be obtained by using
total anaerobic spores as opposed to total aerobic spores. Furthermore, since raw and finished
anaerobic spore counts are typically much lower than aerobic spore counts (Nieminski and
Bellamy 2000), the mathematical difficulties described earlier with low aerobic spore counts
would be worse using anaerobic spore data.

        Another possible Cryptosporidium surrogate is polystyrene (latex) microspheres. These
have been used successfully in pilot-scale challenge studies (e.g., Huck et al. 2002) and have
been proposed for full-scale studies. Under appropriate circumstances, these could be useful
DOP studies, though probably they will be most useful in pilot-scale studies. However, even if
used in pilot-scale studies in place of inactivated Cryptosporidium oocysts, studies will still need
to be conducted using spores to establish similar treatment capability of full-scale and pilot-scale

Raw and finished water spore occurrence
        Table 4.1 summarizes median raw and finished water spore concentrations at the utilities
mentioned earlier using historical data (typically analyzed with a 5 spore/L detection limit (DL)
in finished water) and data collected during the 2003 AWWA study (typically 1 spore/L DL in

March 31, 2009                                                                        Page 53 of 185
finished water). These data show higher raw water spore occurrence in the four river sources
than in the two lake sources, but typically <5 spores/L in finished water from all sources.
Finished water spore levels at all participating facilities were <1 spore/L part of the time (10 to
56 percent of values), especially facilities “A”, “B”, and “E”, indicating that these facilities could
potentially demonstrate lower finished water spore levels if they used a lower DL.

       Since log removal for a process, or part of a process, is calculated by taking the base-10
logarithm of the ratio of the influent and effluent concentrations, as shown in equation 4.1, the
use of lower detection limits for finished water decreases the denominator, thereby increasing the
numerical value of the calculated log removal from raw through filtration. Conversely, use of a
finished water detection limit that is not low enough will result in an under-estimation of the log
removal capability. Therefore, these data demonstrate that in order to mathematically
demonstrate the true treatment capability of many existing facilities, sample volumes >200 mL
will be needed in order to achieve DL of 5 spores/L, and some facilities may need to use sample
volumes >1 L in order to produce DL below 1 spore/L.

Equation 4.1 Calculation of log removal

                                    ⎛ in ⎞
        Log removal = log (base 10) ⎜     ⎟
                                    ⎝ out ⎠

       Figure 4.2 shows the seasonal variation in raw and finished water at one participant, and
also shows the lower values reported for finished water (“filt”) during this study due to the lower
DL used in 2003. This figure shows that raw water occurrence during this study was similar to
the previous historical record, and finished water occurrence also appeared similar, except that
use of lower DL in this study allowed for a more accurate indication of finished water spore
concentration. In fact, at this facility it appeared that an even lower DL could be used since most
of the finished water samples did not have detectable levels of spores using the 1 spore/L DL.

       Table 4.1 indicates that the spore removal capability of facility “B” was >4.5-log during
calendar years 2000 through 2002. During 2003 the same treatment facility was able to
demonstrate >5.4-log spore removal. The increase in the numerical value of the calculated log
removal was due to three factors: a) higher raw water occurrence, b) slightly better treatment
performance, and c) lower finished water DL during this study. The higher median raw water

March 31, 2009                                                                        Page 54 of 185
occurrence (255,000 versus 175,000 spores/L) and increased treatment performance (as
demonstrated in the greater percentage of finished water results that were less than 5 spores/L
[81 versus 98 percent]), may have resulted in a slight increase in the demonstrated spore removal
during the two time periods. However, 0.7 log of this 0.9 log increase was due to the use of a
finished water DL that was five times lower than in the past (log of 5 equals ~0.7). In fact, since
most of the finished water results (56 percent) are still below the new lower DL, it is possible
that a DL lower than 1 spore/L may be needed in order to mathematically demonstrate the true
spore removal capability of this treatment system. Similar spore removal calculations for the
other participants are also summarized in Table 4.1.

       The reduction of the finished water DL by a factor of 5 allowed facilities “A” and “B” to
mathematically demonstrate a greater capability for spore removal, although a slight
improvement in treatment performance was noted as well (see last column in Table 4.1). To
further illustrate this point, Table 4.2 compares the spore removal calculated during calendar
year 2003 at facility “B” using the 1 spore/L DL versus the calculated removal that would have
resulted if a 5 spore/L DL had been used instead. Facility “E” had no historical data to compare
at the higher DL, and facility “F” used a 0.2 spore/L detection limit before and during this study.

       Facilities “C” and “D” demonstrated >4.5-log spore removal from raw to finished water.
The median finished water spore concentrations at these two facilities were <5 spores/L prior to
2003, and 8.5 and 4.6 spores/L, respectively, during this study (2003). These data indicate that
for much of the year a DL of 5 spores/L may be sufficient to mathematically demonstrate the
spore removal capabilities of these two systems. However, during certain seasons of the year the
monitoring data indicate that a lower DL would be useful. For example, between July 1 and
September 30, 2003, the median finished water spore concentration at the two plants was 2.6 and
1.1 spores/L, respectively, resulting in a median log spore removal during this period of >5.4 log.
Therefore, a 1 spore/L detection limit seems warranted for this facility, although a DL of 5
spores/L could be used part of the year, and a DL lower than 1 spore/L should be considered for
other parts of the year.

       Figure 4.3 is a plot of finished water spore concentration versus Cryptosporidium log
removal measured in microbial challenge studies reported in the literature by Huck et al. (2002),

March 31, 2009                                                                      Page 55 of 185
including data from Southern California and Central Canada, and by Dugan et al. (2001),
including studies conducted under normal (optimal) and suboptimal coagulation conditions.
These data suggest that when filters are operated to achieve a level of performance where
finished water spore levels are below 100 spores/L, the facility is capable of at least 3-log
Cryptosporidium removal. Most filtered water samples, including data reported by all
participating facilities in this project, are expected to be well below this level. Data in Figure 4.3
also suggest that when filters achieve performance capable of maintaining finished water spore
levels below 10 spores/L, the filtration facility is capable of >4-log Cryptosporidium removal.
This implies that in the future it may be possible to establish at least a 4-log credit for facilities
achieving <10 spores/L in finished water, or perhaps 5 spores/L as a factor of safety, as an
alternative to calculating log removal of spores as an indicator of Cryptosporidium removal.
Unfortunately, the existing data are not sufficient to justify the credit based on filtered water
concentration alone, although utilities should be encouraged to perform spore analyses as more
sensitive and direct indicators of Cryptosporidium removal than are turbidity and particle counts,
and achievement of the finished water spore levels outlined above should be considered as
treatment objectives.

       Existing historical data coupled with data from this study indicate that some facilities
have filtered water concentrations that are lower than 1 spore/L much of the time, but in general
it appears that even well run filtration plants will detect spores at the 1-5 spore/L level.
Consequently, with a finished water spore concentration between 1-5 spores/L, the raw water
levels will need to be >10,000-50,000 spores/L in order to demonstrate 4-log removal. In the
Draft Rule and Guidance Manual (USEPA 2003), USEPA claims the DOP credit using full-scale
spore monitoring data may not be useful to many utilities because they will not have high enough
raw water spore levels to demonstrate 4-log or greater spore removal. This may be correct for a
number of surface water sources, but may not reflect the situation at the subset of these facilities
that end up in bin 2 or higher that may be able to use and benefit from the DOP credit.

       For example, Nieminski and Bellamy (2000) performed an extensive survey of spore
occurrence in the US which showed that roughly 60% of the facilities had a mean and median
raw water concentration >10,000 spores/L, more than 25% were >50,000 spores/L, and more
than 10 % had >100,000 spores/L (enough for a potential 5-log DOP credit). Consequently, it

March 31, 2009                                                                        Page 56 of 185
appears that full-scale aerobic spore monitoring may be suitable for demonstrating 4.0-log or
greater removal in a large number of US facilities. Furthermore, although there has been no
demonstrated correlation between Cryptosporidium and spore occurrence in raw water (perhaps
due mostly to inadequacies of the Cryptosporidium analytical method), it is possible that the
facilities that happen to be in bin 2 and higher may also include a disproportionate number of
facilities with high raw water spore levels, and consequently more able to gain a benefit from
using the DOP credit.

       In the AWWA and utility sponsored DOP study (Cornwell and Brown [N.d]), three of the
river sources sampled during 2003 had median raw water levels >100,000 spores/L and the other
river source had >40,000 spores/L. The participants have indicated that they expect these
facilities to be in bin 1, but if not they and other river sources with similar raw water spore levels
should be able to mathematically demonstrate at least a 4-log DOP credit if the filters provide the
expected level of treatment (1-5 spores/L). By contrast, one of the lake sources in this study had
a median raw water occurrence of <6,000 spores/L and the other was <400 spores/L. These two
sources are expected to have raw water Cryptosporidium levels well below the bin 2 threshold
(0.075 oocysts/L) and therefore not in need of a DOP credit. Other facilities with similar raw
water spore levels are also not expected to have raw water Cryptosporidium levels high enough
to put them outside bin 1. If these facilities do end up in bin 2 or higher, or if they are in bin 1 but
still want to evaluate the true capability of their treatment process for Cryptosporidium removal,
they probably will not be able to establish this capability using full-scale spore monitoring since
the ambient levels of spores are so low, but could demonstrate this treatment capability using
pilot-scale microbial challenge studies.

Calculating the Numerical Value of the DOP Credit Using Spore Data (mean versus median)
       The following paragraphs discuss alternative approaches to calculate the numerical value
of the DOP credit using spore data from the participating utilities. These alternatives are
described mathematically in equations 4.2 through 4.7.

March 31, 2009                                                                         Page 57 of 185
Equation 4.2 Calculation of mean log spore removal as defined in this report
                                                              ⎡ 1 n             ⎤
                                    ⎛ mean of raw ⎞           ⎢ n • ∑ Raw i ⎥
"log of mean"                  = log⎜                 ⎟ = log ⎢        i =1
                                    ⎝ mean of treated ⎠       ⎢ 1 n             ⎥
                                                              ⎢ n • ∑ Treated i ⎥
                                                              ⎣     i =1        ⎦
                                 ⎡1        n
                                                 ⎤     ⎡1         n
                               = ⎢ • log ∑ Raw i ⎥ −   ⎢n • log ∑ Treated i ⎥
                                 ⎣n      i =1    ⎦     ⎣        i =1        ⎦

Equation 4.3 Calculation of median log spore removal as defined in this report

                                    ⎛ median of raw ⎞
"log of median"                = log⎜                   ⎟
                                    ⎝ median of treated ⎠

Equation 4.4 Calculation of geometric mean log spore removal as defined in this chapter

                                    ⎛ geometric mean of raw ⎞
"log of geometric mean"        = log⎜                           ⎟
                                    ⎝ geometric mean of treated ⎠
                                   ⎡ 1 n            ⎤
                                   ⎢ n • ∏ Raw i ⎥
                             = log ⎢      i =1
                                     1 n
                                   ⎢ • ∏ Treated i ⎥
                                   ⎣ n i=1          ⎦
                               ⎡1 n               ⎤ ⎡1 n                     ⎤
                             = ⎢ • ∑ log (Raw i ) ⎥ − ⎢ • ∑ log (Treated i ) ⎥
                               ⎣ n i =1           ⎦ ⎣ n i =1                 ⎦
Equation 4.5 Calculation of the log ratio for each pair of raw and treated water data used to
                calculate “median of logs” and “mean of logs” as defined in this chapter

                                    ⎛ Raw i ⎞
log ratio i                    = log⎜           ⎟
                                    ⎝ Treated i ⎠

Equation 4.6 Calculation of “median of logs” as defined in this chapter
" median of logs"              = median of log ratio i for all paired data

Equation 4.7 Calculation of “mean of logs” as defined in this chapter (equivalent to log of
                geometric mean of paired data)

                                1 n               1 n     ⎛ Raw i ⎞
" mean of logs"                = • ∑ log ratio i = • ∑ log⎜           ⎟
                                n i =1            n i =1  ⎝ Treated i ⎠
                                 ⎡1 n            ⎤ ⎡1 n                  ⎤
                               = ⎢ • ∑ log Raw i ⎥ − ⎢ • ∑ log Treated i ⎥
                                 ⎣ n i =1        ⎦ ⎣ n i =1              ⎦
                               = "log of geometric mean" as defined in equation 4.4

March 31, 2009                                                                      Page 58 of 185
       Figure 4.4 compares log spore removal at the six participants listed in Table 4.1 using the
five approaches listed above. Please note that although five approaches are listed, two of the
alternatives are theoretically identical (“log of geometric mean” and “mean of logs”), and this is
reflected in the calculated results for the six participants (note identical third and fourth bars in
Figure 4.4 for each participant). Consequently, there are actually only four alternatives, but one
of the alternatives can be mathematically calculated in two separate ways. The numerical value
of the DOP credit listed in Table 4.1 was estimated using the “log of median” approach as
described above. This approach is used throughout this report and is proposed by the authors of
this report as the method to calculate the numerical value of the DOP credit because it is simple
to calculate and produces a calculated value similar to the other calculation methods. However,
the other approach using median (“median of logs”) would be equally suitable, as well as the two
methods related to the geometric mean (“log of geometric mean” and “mean of logs”). In fact
only the “log of means” approach (equation 4.2) is clearly unsuitable for determination of the
DOP credit, as will be demonstrated later in this chapter. Unfortunately, this was the method
selected by USEPA in the draft guidance manual to calculate the DOP credit.

       Two factors that can lead to unsuitability of the “log of means” approach favored by
USEPA for establishing the numerical value of the DOP credit include: a) impact of large
amounts of censored (below DL) data for finished water samples and b) the sensitivity of the
calculated mean raw and treated water concentration when data outliers are present for either
location. Each of these issues will be discussed in more detail below. Another topic to be
discussed below is the issue of whether weekly paired samples of raw and finished water
samples are sufficient, or whether the credit should be calculated using greater sample collection
frequency. A general topic of importance to whatever method is used to calculate the numerical
value of the DOP credit is a recommendation by the authors of this report that this credit only be
calculated using paired data. Finally, another topic specific to the calculation of geometric mean
is included due to numerical limitations of some simple spreadsheet software to calculate
geometric mean for a large number of observations with large numerical values. The following
paragraphs discuss the paired data issue first, then data outliers, censored data, sample collection
frequency, and finally the numerical limitations of some software programs to calculate
geometric mean.

March 31, 2009                                                                       Page 59 of 185
        Recommendation for use of only paired data

        Whatever method is used, approaches that use paired data are probably the most
appropriate since these approaches offer a better opportunity for links that exists between raw
and treated water spore levels, if any, to be reflected. Therefore, “mean of logs” or “median of
logs” would be the most preferable approaches. If any of the other three alternatives are proposed
by USEPA as a method to calculate the DOP credit, these methods should recommend that only
paired data be used to calculate the central tendency (i.e., suggest discarding raw data if treated
result is missing, or vice versa).

        Impact of outliers

        Either of the median approaches described above are relatively insensitive to data
“outliers”, for example incorrect or abnormally high or low numerical values for either raw or
finished water relative to the other data. By contrast, the mean concentration and consequently
the calculated mean log removal using equation 4.2 (“log of means”) can be greatly impacted by
the presence of even a single raw or finished outlier. Table 4.3 and Figures 4.5 and 4.6 illustrate
an example of a situation where one abnormally high finished water sample biased the estimate
of mean log spore removal at one participant during one year. The table and corresponding
figures represent data for calendar year 2001 at facility “B”. Daily samples were collected and
analyzed at this facility for raw, finished, and intermediate points in the treatment process. On
one of the sample collection dates there appears to have been a transcription error. The raw water
result seems correct, but it appears that the pre-sedimentation data value (53,000 spores/L) was
entered into the finished water column by mistake, and all other data for that date were shifted by
one column.

        Table 4.3 and Figures 4.5 and 4.6 illustrate that including this erroneous filtered water
value causes the calculated mean finished water concentration to be >10 times higher, resulting
in a corresponding decrease of >1 log in the log credit calculated using the mean raw and
finished water concentrations [i.e., log(661,868/159) = 3.6 including outlier versus
log(661,868/10.1) = 4.8 without outlier]. By contrast, the median value is less sensitive to
potential outliers and results in the same calculated credit with or without the outlier included
[log(370,000/5) = 4.8 with or without outlier included].

March 31, 2009                                                                     Page 60 of 185
           The “median of logs” approach described earlier is an equally suitable approach to
estimate the log credit. Figure 4.6 includes a plot of the distribution of the paired log removal
values calculated in this manner, and how these relate to “mean” and “median” estimates
calculated using mean and median concentration for the year, as described previously. Using
“median of logs” makes more intuitive sense than the “median” as described in this article
because the “median of logs” maintains the link between the paired raw and finished water
values. Therefore, impacts of raw water conditions on finished water, if any, are reflected in the
calculated values. The “median” approach divorces the two paired values from one another.
However, the “median” approach is simpler and more straightforward to calculate and explain,
and results from the participating utilities indicate little difference between “median” and
“median of logs”, and both are preferable to the “log of means” for reasons described in this

           In this particular example, the log credit should be at least 4.8, and this result can be
mathematically demonstrated using either of the approaches described above using the median,
or using the mean if the probable outlier (53,000 spores/L) is excluded. In fact, it appears that
there may be more than just one potential outlier biasing the mean (more than 95 percent of the
finished water values are lower than the mean in Figure 4.5, with or without excluding the one
identified potential outlier). Therefore, if the “log of means” approach outlined in equation 4.2 is
used to determine the numerical value of the DOP credit as described above, a protocol needs to
be incorporated to identify and account for and exclude outliers. In this example, the outlier was
a finished water value that was too high, resulting in an under-estimate of the DOP credit. There
were examples from other participants in which a single high raw water value resulted in an
over-estimate of the DOP credit using the mean approach, unless this outlier was excluded. A
better and simpler alternative to using the mean and then developing and employing a
complicated procedure to evaluate and identify potential outliers is to use the median approach
described above. The median is simple and straightforward to determine and yet would be less
sensitive to these potential outliers.

           Impact of censored data

           Inspection of Figure 4.5 also reveals another reason to use median as opposed to mean to
calculate the numerical value of the DOP credit. More than 85 percent of the finished water

March 31, 2009                                                                       Page 61 of 185
values are below the 5 spore/L DL used at this facility during calendar year 2001. Greater than
55 percent of the finished water concentrations were less than the DL in 2003, even though the
DL was lowered to 1 spore/L (see Table 4.2).

       When calculating the mean or determining the median of data with censored values, one
approach is to substitute the numerical value of the DL for all of the censored data. Other
methods, such as using half the DL, are often used but are far too arbitrary. A method to estimate
descriptive statistics for flood discharge data containing censored values is described by Helsel
and Cohn (1988) and Hirsch and Stedinger (1987). The method (“plotting positions”) essentially
involves predicting the distribution of the censored data by evaluating the distribution of the
uncensored (i.e., the above DL) data. Figure 4.7 compares the actual data from 2003 for facility
“B” using the DL value for censored data versus synthetic data generated using the “plotting
positions” method described in the two articles referenced above. Please note that 56 percent of
the finished water values are censored, plus spores were detected in an additional 18 percent of
the samples at exactly 1 spore/L (the numerical value of the DL).

       This analysis of the data reveals that the predicted median using the synthetic data is
about 0.23 spores/L, which would result in a calculated log spore removal of 6.0-log, or 0.6-log
higher than estimated using 1 spore/L (i.e., the DL) as the median. The calculations for the mean
show the impact of two factors. First, the mean using the synthetic data is slightly lower (about
7.2 versus 7.7 spores/L), but the mean in either case represents about the 93 percentile of the
data, indicating that the mean may be biased by some outliers in the upper 7 percent of the data.
Consequently, the mean can be biased by data outliers, as noted earlier, and this bias can be
further exacerbated by using the value of the DL for the censored data when calculating the mean
or other descriptive statistics. Use of lower detection limits and employing techniques to estimate
the distribution of censored data can help better estimate the mean, but a method to identify and
account for data outliers is also needed. By contrast, a simpler and better estimate of the central
tendency of the data can be obtained without employing complex methods to identify and
account for outliers or censored data by: a) using the median instead of the mean and b) using a
detection limit low enough so that more than 50 percent of the finished water values are above
the DL. Even if more than 50 percent of the values are below the DL, using the value of the DL
as the median will still be better than using the mean because of the potential impact of data

March 31, 2009                                                                     Page 62 of 185
outliers on the calculation of the mean. Some facilities may wish to choose to employ more
complicated methods to account for censored data, like the approach using “plotting positions”
described earlier, to better estimate the median, though it should not be a requirement for all

        Weekly versus more frequent spore monitoring to establish the DOP credit

        Section of the MTGM indicates that calculating DOP credit using spore
monitoring data should be based on sampling frequency of at least once per week for 52
consecutive weeks. More frequent collection is probably not necessary, though utilities can
voluntarily choose to perform more frequent monitoring. For example, one utility has been
collecting daily spore samples from multiple locations in the treatment process, including raw
and finished water, since January 2000. The facility uses this data to evaluate treatment
performance on a daily basis. Table 4.4 compares the results for median log removal from raw to
finished water at facility “B” using daily paired samples collected during calendar year 2001
versus similar calculated median log spore removal using data collected once per week. This
comparison shows that the numerical results are not much different using daily samples versus
any of the sets of weekly data. Therefore, while some facilities may wish to collect samples more
frequently than once per week for their own purposes, it should not be necessary to collect
samples more frequently than once per week to establish the DOP credit.

        Numerical limitations of some software to calculate geometric mean

        Previous discussion indicates that “log of means” is not an appropriate approach for
calculating the numerical value of the DOP credit. For this reason, either of the approaches using
the median are preferable because they are simple and straightforward to determine and yet are
less sensitive to data outliers than “log of means” that USEPA prefers.

        However, USEPA may have some interest in replacing the mean with geometric mean,
instead of using median, in particular because the geometric mean is also less sensitive to data
outliers. For example, the numerical value of the DOP credit during 2003 for the six facilities in
Table 4.1 using the median and geometric mean are similar (median and geometric mean
estimates of the DOP credit for these facilities are as follows: “A” = 4.2 vs. 4.0, “B” = 5.4 vs.
5.3, “C” = 4.7 vs. 4.9, “D” = 4.9 vs. 4.9, “E” = 3.3 vs. 3.1, and “F” = 2.1 vs. 2.0). Similarly,

March 31, 2009                                                                     Page 63 of 185
conclusions for data at Facility “B” in Table 4.4 would be similar if median or geometric mean
were used (DOP credit ranges from 4.5 to 4.6 using either median or geometric mean, and about
4.55 using data from all seven days of the week). The geometric mean can also benefit from use
of methods to account for censored data. The estimate of geometric mean of filtered water
concentration for the synthetic data in Figure 4.7 is 0.23 spores/L, the same as the estimate using
the median. Since the geometric mean of the raw water concentration is slightly higher than the
median of raw water, the calculated value for the DOP credit using geometric mean is slightly
higher (6.1 vs. 6.0).

        However, the “log of geometric mean”, though theoretically suitable, may create some
difficulties during implementation for utilities depending on the geometric mean function in
some commonly used spreadsheet software. For example, the geometric mean function used in
Microsoft EXCEL 2000 (“geomean”) has a limitation of about 150 data values when the
geometric mean has a numerical value of about 0.01, and a limitation of 70-80 data points when
the numerical value of the geometric mean is about 10,000, such as you might find in a raw
water for spores expressed in units of spores/L. By contrast, some other software may not suffer
from this limitation. For example, the geometric mean function (“@geomean”) in the Lotus
Development Corporation spreadsheet software called “Lotus 1-2-3” does not have this

        Consequently, although most utilities using the DOP credit should not have any difficulty
since they will typically only have about 50 or so data points (weekly paired data for one year),
utilities that want to use more than one year of historical data or that are collecting samples more
than once a week may have some difficulty due to this numerical limitation in some spreadsheet
software. They can get around this limitation, when it exists, by manually calculating the
geometric mean for raw and treated water without using the geometric mean function. This can
be accomplished by taking the mean of the base-10 log values, then taking the anti-log of the
mean of these log values as shown in equation 4.8. Furthermore, as shown in Equation 4.7, a
mathematically equivalent calculation of the “log of geometric mean” can be determined by
using the “mean of logs” approach.

March 31, 2009                                                                      Page 64 of 185
Equation 4.8 Calculation of geometric mean when numerical limitations do not allow direct
              calculation using internal spreadsheet functions (e.g., “geomean” function in
              Microsoft EXCEL)
                                                 ⎧1 n             ⎫
                                                 ⎨ • ∑ log (x i ) ⎬
geometric mean of " x"                    = 10   ⎩ n i=1          ⎭
                                                                      = 10mean of "log xi "

       Summary of methods to determine numerical value of DOP credit

       Two alternatives related to median and two alternatives related to the geometric mean are
the preferred methods for establishing the numerical value of the DOP credit using spore data or
other Cryptosporidium surrogates. Of these, either approach using median are simplest and most
straightforward to determine. The two approaches related to geometric mean (see equations 4.4
or 4.7) are also suitable, though the “log of geometric mean” approach as outlined in equation
4.4 may be complicated under certain situations described above. The “log of means” approach
outlined in equation 4.2 is identical to the approach proposed by USEPA in the Microbial
Toolbox Guidance Manual. This approach is the only approach described above that can be
demonstrated as clearly unsuitable for use in establishing the numerical value of the DOP credit,
unless complicated provisions are included by USEPA to help identify and manage data outliers
and censored data. Instead of requiring development of complicated guidance needed for other
approaches, USEPA should explicitly recommend either of the median approaches to establish
the numerical value of the DOP credit. If these other approaches are also listed, they should be
described as alternatives to the median approaches and the guidance manual should adequately
outline the limitations of these other approaches so states and utilities reviewing the guidance
manual can make informed decisions.

       Whatever method is adopted, use of data collected at least once a week for one year
should be sufficient to establish the numerical value of the DOP credit. Sample collection for
periods longer than one year or at frequencies greater than once a week should not be required by
the states or USEPA, however utilities that voluntarily choose to collect more data should be
allowed to do so. Samples should be collected at regular, evenly spaced intervals throughout an
entire 52 week period, or multiples of 52 weeks when more than one year of data is used.

March 31, 2009                                                                            Page 65 of 185
However, utilities should be allowed some discretion on sampling dates and number of samples,
especially if more than 50 samples are collected. For example, 50 out of 52 samples for weekly
sampling dates or about 350 daily samples excluding holidays seems reasonable. In addition,
while utilities collecting samples once a week, for example, should be encouraged to collect
samples on the same day each week, utilities should be allowed some discretion in establishing
sample collection dates as needed due to holidays, vacations, or other issues related to
availability of sample collection personnel.

Using full-scale spore monitoring to evaluate treatment performance other than DOP
       Utilities performing spore monitoring to evaluate performance of treatment facilities may
get some value out of collecting samples from raw water and intermediate points in the process.
However, most utilities will focus most of their attention on finished water spore levels as an
indicator of performance of the filters in particular, and the entire treatment process in general.
For example, since spores are more sensitive than particle counts or turbidity and reflect more
closely factors that may impact Cryptosporidium removal, monitoring of filtered water spore
levels like that illustrated in Figure 4.8 can be used to illustrate improvements over time due to
changes in infrastructure, operational practices, or management strategies. This greater
sensitivity may allow demonstration of treatment enhancements for costly but difficult to
quantify improvements, such as operational practices and management policies. Figure 4.8
shows progressive improvement over time at one facility from 1997 to 2002 due to refinements
in operational practices. Chapter 6 includes spore monitoring data to illustrate treatment
performance during different stages at lime softening plants.

Pilot-Scale Microbial Challenge Studies

       A DOP pilot-scale challenge study would involve measuring removal of spiked
Cryptosporidium or appropriate surrogate in a 2 to 5 gpm pilot facility mimicking the full-scale
treatment plant. Source water for the pilot facility should be provided from the location
representative of where the Cryptosporidium bin assignment samples are/were collected.
Although explicit guidance on pilot studies is not available from USEPA, the following would be
reasonable for most facilities:

March 31, 2009                                                                     Page 66 of 185
          •   Challenge studies conducted for two (2) weeks during each of four consecutive

          •   The study should be preceded by two (2) weeks of side-by-side studies comparing
              performance of full-scale facility to pilot plant operated under identical conditions
              (it must be shown that full-scale performance equals or exceeds pilot-scale
              performance through the final filtration stage)

          •   Monitoring of raw, spiked, and finished water grab samples to include turbidity,
              particle counts, and Cryptosporidium (or other microbial indicator)

          •   On-line, continuous turbidity and particle count monitoring of individual filter
              and combined filter effluent is recommended for the benefit of the utility so they
              can evaluate operations of the pilot facility, but this monitoring should not be a
              requirement or condition for the credit

          •   Pilot-study protocol to be negotiated with, and approved by the State.

          •   Objectives of testing should be to establish performance under routine or typical
              conditions (not “worst-case” scenarios – see later discussion), with “typical”
              performance established by monitoring removal of Cryptosporidium (or approved
              surrogate) in combined filter effluent (not in individual filters – see discussion

          •   DOP credit established in pilot studies will be applicable to full-scale plant at full-
              scale flow rates consistent with unit process loading rates used in pilot-scale unit
              processes (flocculation, clarification, filtration) during DOP evaluation, unless
              otherwise negotiated with the State

       The above discussion proposes an aggressive schedule incorporating two weeks of
quarterly studies, although a less aggressive schedule would be appropriate under most
conditions. For example, a few weeks of one-time testing would be appropriate for many
facilities, and would be consistent with existing performance testing requirements for some
states. However, Chapter 12 of the draft MTGM outlines an extravagantly expensive and overly

March 31, 2009                                                                       Page 67 of 185
aggressive program that includes 52 weeks of testing (Table 12.3, page 12-13 of drat MTGM).
Furthermore, the objective of testing should be to represent performance under routine or typical
conditions, not “worst-case” conditions emphasized in the MTGM. This is consistent with the
characteristics of the LT2ESWTR framework which incorporates typical Cryptosporidium
occurrence, viability, and infectivity along with typical treatment performance to establish
overall risk to drinking water consumer. In addition, “typical” treatment performance is reflected
by measuring treatment performance from raw (plus spike) through combined filter effluent, not
raw through each individual pilot filter. This is consistent with the requirements for the DOP
credit using spores, and is also consistent with Section of the MTGM, although Table
12.3 on the previous page of the guidance manual implies the opposite.

        The DOP credit established in pilot studies should be applicable to the full-scale plant at
full-scale flow rates consistent with unit process loading rates used in pilot-scale unit processes
(flocculation, clarification, filtration) during the DOP evaluation, unless otherwise negotiated
with the State. For example, a pilot study which achieves >4.5 log removal of Cryptosporidium
using the State approved protocol when pilot filters were operated at 8 gpm/sf means that the
full-scale plant should be certified for a total credit of at least 4.5, not the 3.0 automatic credit, as
long as the full-scale plant filtration rate is <8 gpm/sf. This does not mean the State is required to
certify the full-scale plant for filtration rates up to 8 gpm/sf. However, it does mean that as long
as the State establishes the maximum filtration rate for the plant anywhere below 8 gpm/sf (e.g.,
4.5 gpm/sf), the State can not award a credit lower than 4.5.

        Spore monitoring results can also be used to evaluate performance of pilot-scale
facilities, in particular comparison of pilot- versus full-scale performance, like results shown in
Figure 4.9. Most water treatment professionals with experience interpreting pilot study data have
observed similar results numerous times demonstrating that pilot-scale facilities do not
overestimate full-scale treatment performance in areas such as removal of spores and other
particulates. In fact, results like those in Figure 4.9 demonstrating better spore or particulate
removal in full-scale clarifiers and filters than in analogous pilot-scale facilities are commonly

March 31, 2009                                                                          Page 68 of 185
Other DOP issues

        Some DOP issues were inadequately addressed or not addressed at all in the Draft
Toolbox Guidance Manual for the LT2ESWTR. These issues are discussed separately below:
costs, maintenance of toolbox credits, process specific studies, and potential DOP penalty.

        The cost for the DOP credit using either aerobic spores or pilot-scale microbial challenge
studies would not be dependent upon the size of the facility being demonstrated. For example, a
full-scale spore monitoring study will include the same number of samples collected at the same
sampling frequency whether a facility is 2 mgd or 200 mgd (one raw and one finished water
sample, once per week for a year is proposed for both). A pilot-scale facility will similarly be
about the same size (~5 gpm) and will include about the same duration of testing (about 2 weeks
of testing per quarter for one year is reasonable) associated with full-scale facilities of a variety
of sizes (the one difference perhaps is that larger utilities may have enough in-house staff to
design, build, and operate the pilot whereas a smaller utility may not be able to do it without
hiring an outside consultant). Therefore, the DOP credit will be cost-effective for a wide variety
of facility sizes, and will be even more cost-effective for larger facilities.

        A DOP spore study is expected to involve a minimum of 50 weeks of two paired spore
samples (plant influent and effluent) per week. Median cost for spore analysis from a survey of
eight commercial laboratories is about $60/sample. The resulting analytical cost is about $6,000
for the entire study. Cost for data interpretation and report preparation, would increase the total
cost to about $60,000 for the entire study. Facilities performing the analyses in-house could
reduce costs even further. Many facilities may already possess most of the required equipment
(an incubator, an autoclave, a membrane filtration apparatus, and a microscope), and may only
need to purchase a shaking water bath capable of achieving and maintaining at least 90°C
($3,000 to $6,000) or a stirring hotplate ($300-$1,200) capable of achieving the same
temperatures. The analytical technique does not require special training or certification, for
example like is needed for Cryptosporidium analyses, for personnel familiar with standard
microbial analyses, in particular the membrane filtration technique for coliform analyses. For a
pilot-scale study, a 5-gpm pilot plant with all required hardware and instrumentation, plus

March 31, 2009                                                                       Page 69 of 185
design, construction, operation, monitoring, data evaluation, and report preparation would
require $600,000 or less. Consequently, even the latter pilot-scale DOP cost is more cost
effective than some other toolbox credits for facilities capable of achieving up to 4.0 log of
Cryptosporidium removal (or approved surrogates), and becomes even more cost effective for
larger facilities. The guidance manual should clearly indicate that spores are an accepted
surrogate for Cryptosporidium spiking.

Maintenance of DOP credit
       Once established, the DOP credit should be retained as long as the facility maintains
compliance with the IESWTR. Determination of alternative compliance criteria, and
requirements to comply with these alternate criteria, are not appropriate. The DOP credit
establishes a correction or adjustment to the Cryptosporidium protection capability of an existing
treatment facility in compliance with the IESWTR. Whereas a utility in compliance with the
IESWTR is allowed a minimum 3.0 credit (2.5 for direct filtration) without having to prove what
the true treatment capability of the existing system really is, once a utility completes a DOP
study and proves what the actual capability of the system is when it is in compliance with the
IESWTR, it should not be required to establish and attain a compliance requirement beyond
IESWTR compliance.

Process specific demonstration studies
       DOP studies can be applied to entire treatment processes (raw to finished), to
intermediate segments of the treatment process (e.g., pre-sedimentation effluent to finished
water), or to individual treatment processes (second stage of two filter process). Due to practical
limitations resulting from decreased ability to mathematically demonstrate DOP credit when
process influent levels are lower, the DOP approach using ambient spores becomes less useful
when the starting point for the process gets later and later in the process. Consequently, DOP
studies using full-scale spore monitoring work best when measured from raw water to some later
point in the treatment process. Pilot-scale challenge studies can be used instead of full-scale
spore monitoring studies under conditions when influent ambient spore levels are too low,
whether this is due to low raw water levels or due to low ambient influent levels at points in the
process after previous stages in treatment.

March 31, 2009                                                                     Page 70 of 185
       One unit process that is potentially particularly well suited for DOP studies using spore
monitoring is RBF. DOP studies for RBF systems can be performed at full-scale pilot facilities,
for example a monitoring well/collector representing a future monitoring well network or other
river bank, river bottom, lake bank, or lake bottom collection system similar to the pilot
well/collector. This facility will not be able to use microbial challenge studies, but can measure
removal of ambient Cryptosporidium indicators, such as aerobic spores. DOP studies for RBF
are unique in that these are the only opportunities to use DOP to establish UBT credits.
Therefore, a facility that is ineligible for an automatic RBF credit for one reason or another
(media gradation, distance from surface water source, etc.) can receive whatever credit can be
proven in a DOP study, and the first 1.0 of this credit can be used for UBT and the remainder can
be used for the additional 1.0 or 1.5 credits needed for bins 3 and 4, respectively. Once the pilot
well/collector establishes the Cryptosporidium protection capability of the riverbank in the
vicinity of the well/collector, the utility will be allowed to construct the RBF system outlined in
the State approved DOP protocol and the facility will be awarded the credit demonstrated in the
pilot well without need of further testing (analogous to the alternate intake indirect “credit” in the
toolbox). This credit will be useable until the second round of Cryptosporidium bin assignment
samples, at which time the new bin assignment sample location will be the combined discharge
from the RBF system.

       Figure 4.10 and Table 4.5 summarize spore removal in a pair of wells associated with an
existing RBF system. Mean Cryptosporidium occurrence in the river source is >0.8 oocysts/L,
which is close to the bin 3 threshold of 1.0 oocysts/L. Giardia occurrence in the river is also
high. Yet, no Giardia or Cryptosporidium have been detected in the RBF wells. Spore
monitoring data indicates that the spore removal capability for the RBF process at this location is
at least 4.0 log. Since the wells are existing, this RBF system is not eligible for a direct RBF
credit, although it will receive what amounts to an indirect credit of 1.0 credit since the river
sample would put the facility in bin 2, at minimum, but collecting bin assignment sample from
the RBF system, as mandated by the LT2ESWTR, will ensure assignment to bin 1 at this facility.
However, if these were pilot wells for a proposed RBF system, results from spore monitoring
would justify a credit >2.5 log (including 1.0 credit of UBT), sufficient for any bin assignment

March 31, 2009                                                                        Page 71 of 185
Potential DOP penalty
        DOP studies inherently provide conservative estimates of the true treatment performance
of the full-scale facilities being evaluated. For example, the DOP credit established using full-
scale aerobic spore monitoring includes an inherent safety factor because the numerical value of
the Cryptosporidium removal credit will be assumed to be equal to the measured spore removal,
even though Cryptosporidium removal in literature studies is always greater than spore removal
(see later discussion). Similarly, pilot-scale microbial challenge studies will be conservative
indicators of full-scale Cryptosporidium removal (especially if conservative Cryptosporidium
indicators like spores are used) because widespread experience of water treatment practitioners
with experience using pilot studies indicates that full-scale clarification and filtration facilities
routinely provide greater removal of particulates, turbidity, and microbial indicators than do
pilot-scale systems. In addition, even if this was not a common observation for pilot facilities, the
USEPA requirements for the DOP credit using pilot studies include a precaution requiring the
verification of the relationship between pilot-scale and full-scale facilities associated with the
DOP study by requiring a period of comparison where both are operated under identical
conditions (same influent raw water, similar coagulation conditions, filter and clarifier loading
rates, etc.).

        A provision in the draft rule (section IV.C.17) and in Chapter 12 of the microbial toolbox
guidance manual allows State’s the discretion to potentially penalize facilities that try a DOP
study and end up mathematically demonstrating a total credit that is less than the automatic
credits allowed by the Rule. Therefore, for example, a State can choose to award a 3.3 credit to a
facility that qualifies for 3.5 total automatic credits if the utility was only able to mathematically
demonstrate 3.3 log removal in a DOP spore study or pilot study.

        However, it is difficult to see any benefit from inclusion of this potential penalty since the
automatic credit is based on a national mean, not on individual performance. The use of national
mean was purposely done by USEPA and deducting credits at an individual plant basis is
contrary to that approach. Several huge potential negative consequences are associated with
inclusion of this penalty provision, including the potential to increase costs and nuisance for
facilities attempting to achieve the DOP credit using spores, and the potential to deter facilities
that may have benefited from using the credit from even trying for the credit.

March 31, 2009                                                                        Page 72 of 185
        In summary, the prospective penalty USEPA is attempting to incorporate as part of the
DOP credit requirements demonstrates a lack of understanding of the conservative nature of
proposed demonstration studies for the DOP credit (full-scale spore monitoring or pilot-scale
microbial challenge studies). This potential penalty is not expected to be applied to many, if any,
facilities since most that will try for the credit are expected to demonstrate sufficient removal in
pilot-scale microbial challenge studies or full-scale studies. However, the greatest potential
impact of this potential penalty is that the mere prospect that a utility could be penalized for
trying to achieve a DOP credit will deter a number of utilize from even trying for the credit, even
though most facilities will probably be able to demonstrate great enough Cryptosporidium
removal capabilities that they not only would not be subject to a penalty, but would be able to
easily establish a higher credit for their existing facilities using the DOP credit. Therefore, a
number of utilities may pursue more expensive or less readily achievable credits that could have
been achieved more easily and inexpensively using the DOP credit.


   1.    Literature data indicates that treatment facilities capable of achieving finished water
         turbidities <0.3 ntu as needed for compliance with the IESWTR and LT1ESWTR can
         achieve >4-log removal of Cryptosporidium. A large portion of the credits available to
         utilities in the LT2ESWTR Microbial Toolbox, including the DOP credit, were
         included so that existing facilities capable of achieving this higher level of performance
         could be allowed the credit they deserve.

   2.    Cryptosporidium is typically removed more readily than aerobic spores during physical
         treatment process like clarification and filtration. Consequently, facilities that
         demonstrate around 4-log removal of aerobic spores during treatment are probably
         capable of greater than 4-log Cryptosporidium removal. This is the basis behind
         allowing use of full-scale monitoring of ambient (i.e., background) aerobic spore levels
         in order to calculate the numerical value of the DOP credit.

   3.    Median finished water spore concentration in well run treatment plants typically ranges
         from 1 to 5 spores/L.

March 31, 2009                                                                      Page 73 of 185
   4.    Facilities with 1 to 5 spores/L in finished water will need more than 10,000 to 50,000
         spores/L in raw water in order to demonstrate a 4-log DOP credit. 25 percent of the
         facilities during a national study conducted by Nieminski and Bellamy (2000) had a
         median spore concentration of >50,000 spores/L. Furthermore, it is possible that the
         number of facilities that happen to be in Bin 2 and higher as a result of LT2ESWTR
         required Cryptosporidium monitoring may also include a disproportionate number of
         facilities with high raw water spore levels, and consequently more able to gain a benefit
         from using the DOP credit.

   5.    Full-scale spore studies reported by Cornwell and Brown ([N.d.]) showed that a DOP
         credit greater than 4.5 could be demonstrated at four facilities from river sources with
         raw water spore concentrations >10,000 spores/L.

   6.    Facilities with fewer than 10,000 spores/L in the influent water (see equation 4.1) will
         probably need to use pilot-scale microbial challenge studies to establish the full DOP
         credit these facilities deserve.

   7.    USEPA needs to explicitly acknowledge that monitoring aerobic spore removal in full-
         or pilot-scale studies is a suitable surrogate and provides a conservative estimate of
         Cryptosporidium removal, and consequently spore monitoring is suitable for establish
         the numerical value of the DOP credit.

   8.    Spore monitoring can be used to determine treatment removal for an entire treatment
         process, as well as individual components of the process. For example, participating
         utilities demonstrated >0.8 median log removal associated with adding a second
         filtration stage (post-filter GAC), 0.6 to 1.6 median log removal for pre-sedimentation,
         1.4 median log removal for a SFBW clarifier, and >4 log removal in RBF wells.

   9.    Spore monitoring has been used by water utilities to evaluate treatment performance.

   10.   The alternative to monitoring ambient aerobic spore removal in full-scale treatment
         processes in order to establish the DOP credit is to use pilot-scale microbial challenge
         studies. However, spore monitoring can also be an important part of these studies as
         well, including: a) use of spores instead of Cryptosporidium as the challenge organism

March 31, 2009                                                                    Page 74 of 185
        in the spiking studies and b) to verify similar (or lower) treatment performance in the
        pilot-scale facility versus the full-scale facility when both are operated in a similar
        manner and are treating similar source water. It may also be possible to combine the
        two in a hybrid approach where pilot-scale studies are conducted to establish a
        correction factor relating Cryptosporidium to spore removal at a given facility, and this
        correction factor applied to full-scale ambient aerobic spore monitoring results.

APHA (American Public Health Association) 1993. Aerobic Bacterial Spores (Section 8.9).

       Standard Methods for the Examination of Dairy Products. 16th ed. Edited by Robert T.

       Marshall. Washington, DC: APHA, pp. 280-281.

APHA (American Public Health Association) [N.d.]. Method 9218: Aerobic Endospores in

       Water. Standard Methods for the Examination of Water and Wastewater. 21st ed.

       Washington, DC: APHA, forthcoming.

Clark, S., M. Morabbi, T. Hargy, J. Chandler, and J. Wiginton. 2001. Softening: The Ultimate

       Microbial Tool? In Proc. of 2001 Water Quality Technology Conference. Denver, CO:


Cornwell, D. and R. Brown. [N.d.]. DOP Credit in the LT2ESWTR Using Ambient Spore
       Monitoring. Jour. AWWA. (pending).
Cornwell, D., M. LeChevallier, M. MacPhee, N. McTigue, H. Arora, G. DiGiovanni, and J.

       Taylor. 2000. Treatment Options for Giardia, Cryptosporidium, and other contaminants

       in Recycled Backwash Water. Denver, CO: AwwaRF and AWWA.

Cornwell, D., R. Brown, M. MacPhee, and B. Wichser. 2003a. Application of the LT2ESWTR

       Microbial Toolbox at an Existing Surface Water Treatment Plant. [full article =

       Available:           (November    12,

       2003)]. Expanded summary available at Jour. AWWA 95:9:76-79.

March 31, 2009                                                                    Page 75 of 185
Cornwell, D., M. MacPhee, R. Brown, and S. Via. 2003b. Demonstrating Cryptosporidium

       Removal using Spore Monitoring at Lime Softening Plants. Jour. AWWA., 95:5:124.

Dugan, N., K. Fox, J. Owens, and R. Miltner. 2001. Cryptosporidium Control Through

       Conventional Treatment. Jour. AWWA., 93:12:64.

Emelko, M. 2001. Removal of Cryptosporidium parvum by Granular Media Filtration. Ph.D.

       diss. University of Waterloo, Waterloo, Ontario, Canada.

Hall, T., J. Pressdee, and N. Carrington. 1994. Removal of Cryptosporidium Oocysts by Water

       Treatment Process. London, England: Foundation for Water Research Limited.

Helsel, D., and T. Cohn. 1988. Estimation of Descriptive Statistics for Multiply Censored Water

       Quality Data. Water Resources Research, 24:12:1997-2004.

Hirsch, R., and J. Stedinger. 1987. Plotting Positions for Historical Floods and Their Precision.

       Water Resources Research, 23:4:715-727.

Huck, P., et al. 2001. Filter Operation Effects on Pathogen Passage. Denver, CO: AwwaRF and


Mazounie, P., F. Bernazeau, and P. Alla. 2000. Removal of Cryptosporidium by High Rate

       Contact Filtration: The Performance of the Prospect Water Filtration Plant During the

       Sydney Water Crisis. Water Science and Technology. 41(7):93-101.

McTigue, N., M. LeChevallier, H. Arora, and J. Clancy. 1998. National Assessment of Particle

       Removal by Filtration. Denver, CO: AWWARF and AWWA.

Nieminski, E. and W. Bellamy 2000. Application of Surrogate Measures to Improve Treatment

       Plant Performance. Denver, CO: AwwaRF and AWWA.

Ongerth, J. and J. Pecoraro. 1995. Removing Cryptosporidium Using Multimedia Filters. Jour.

       AWWA, 87(12):83-89.

March 31, 2009                                                                   Page 76 of 185
Patania, N., J. Jacangelo, L. Cummings, A. Wilczak, K. Riley, and J. Oppenheimer. 1995.

       Optimization of Filtration for Cyst Removal. Denver, CO: AwwaRF and AWWA.

Rice, E., K. Fox, R. Miltner, D. Lytle, and C. Johnson. 1996. Evaluating Plant Performance with

       Endospores. Jour. AWWA., 88 (9): 122-130.

States, S., R. Tomko, M. Scheuring, and L. Casson. 2002. Enhanced Coagulation and Removal

       of Cryptosporidium. Jour. AWWA, 94(11):67-77.

Swaim, P., M. Heath, N. Patania, W. Wells, and R. Trussell. 1996. High-rate Direct Filtration for

       Giardia and Cryptosporidium Removal. In Proc. of 1996 Annual Conference. Denver,

       CO: AWWA.

West, W., P. Daniel, P. Meyerhofer, A. DeGraca, S. Leonard, and C.P. Gerba. 1994. Evaluation

       of Cryptosporidium Removal Through High-Rate Filtration. Proceedings of 1994 Annual

       Conference. Denver, CO: AWWA.

USEPA. 2000. Stage 2 Microbial and Disinfection Byproducts Federal Advisory Committee

       Agreement in Principle. Fed. Reg. 65(251):83015-83024.

USEPA. 2003. National Primary Drinking Water Regulations: Long-Term 2 Enhanced Surface

       Water Treatment Rule. Fed. Reg. 68(154):47640-47795.

Yates, R., K. Scott, J. Green, J. Bruno, and R. DeLeon. 1998. Using Aerobic Spores to Evaluate

       Treatment Plant Performance. In Proc. of 1998 Annual Conference. Denver, CO:


March 31, 2009                                                                    Page 77 of 185
                                                 stable - pre-filter spike   stable - raw water spike   non-stable


         Log Spore Removal





                                 0           1                 2                 3               4               5       6
                                                                Log Cryptosporidium Removal

        Sources: Clark et al. 2001, Dugan et al. 2001, Huck et al. 2002, Mazounie et al. 2000, Yates et al. 1998

        Figure 4.1                   Comparison of Cryptosporidium and spore removal in pilot-scale spiking studies

March 31, 2009                                                                                                        Page 78 of 185
                                                                                Table 4.1

                                                  Summary of Median Aerobic Spore Removal at Project Participants

     Facility                  Time               Source   Number of    Median Concentration            Log Removal           Percent of Finished
                               Period             Water    Sampling          (spores/L)               (Raw to Finished)         Water Results
                                                             Dates                                                               <5 spores/L
                                                                          Raw           Finished

         A           1996-2002 (old DL)           River       325        90,000             20.               3.65                    26.3
                       2003 (new DL)                          41         43,000             3.0               4.16                    63.4

         B           2000-2002 (old DL)           River      1,072      175,000             <5               >4.54                    81.4
                       2003 (new DL)                          225       255,000             <1               >5.41                    97.8

         C               1998-2000                River       251       290,000             <5               >4.76                    64.5
                       2003 (new DL)                          67        435,000             8.5              4.71                     41.2

         D           2001-2002 (old DL)           River       287       300,000             <5               >4.78                    33.1
                       2003 (new DL)                          64        377.778             4.6              4.92                     51.6

         E             2003 (new DL)               Lake        17         5,450             3.0               3.26                    70.6

         F                1997-2001                Lake       213          335              9.9               1.53                    21.7
                          2002-2003                           79           355              2.8               2.10                    64.6

Facility descriptions:
                A= coagulation, flocculation, clarification, sand filtration, GAC filtration
                 B= two-stage lime softening preceded by pre-sedimentation (no "coagulant", other than lime)
           C & D= two-stage lime softening preceded by pre-sedimentation ("coagulant" added along with lime)
                 E= coagulation, flocculation, clarification, sand filtration
                 F= direct filtration

         DL= detection limit            new DL = 1 spore/L - used during 2003       old DL = 5 spore/L - historical data for most facilities

March 31, 2009                                                                                                                                      Page 79 of 185
                                                                          Table 4.2

Comparison of mathematically demonstrated log removal that would have resulted at Facility “B” if a 5 spore/L detection limit (DL) had been used instead of a 1
                                                                 spore/L detection limit

                              Finished        Median Concentration               Log Removal            Percent of Finished Water

                              Water DL         Raw           Finished          (Raw to Finished)               Results <DL

                              (spore/L)      (spore/L)      (spore/L)

                                 1            255,000           <1                    >5.41                        55.8

                                 5                              <5                    >4.71                        97.8

        DL= detection limit

March 31, 2009                                                                                                                               Page 80 of 185
                                                                               Table 4.3

Comparison of mathematically demonstrated log removal that would have resulted at Facility “B” if a 5 spore/L detection limit (DL) had been used instead of a 1
                                                                 spore/L detection limit

     Value                                           Concentration                                                        Log Spore Removal

                                                       (spores/L)                                                          (Raw to Finished)

                           Raw                                      Finished                               Outlier included               Outlier excluded

                                             Outlier included                  Outlier excluded

     Mean                 661,868                    159                             10.1                         3.62                          4.82

    Median                370,000                    <5                               <5                          4.87                          4.87

Median of paired data †                                                                                           4.87                          4.87

†   =   Calculate log of the ratio of each pair of raw versus finished water data, then determine median of these values – see Figure 4.5. This value is called

        “median of logs” elsewhere in this report.

‡   =   Probable outlier due to data entry error (pre-sedimentation basin effluent sample result entered incorrectly as finished water) – value of probable outlier

        was 53,000 spores/L

March 31, 2009                                                                                                                                  Page 81 of 185
                                                   Table 4.4

 Comparison of calculated log spore removal using daily spore samples for a year versus samples collected once a
                                 week during calendar year 2002 at facility “B”

 Day of the Week                Median                  Median Log Removal                    Percent of

                        Concentration (spores/L)                                       Finished Water Results

                          Raw           Finished                                             <5 spores/L

Sunday                  210,000            <5                   >4.62                           92.0

Monday                  180,000            <5                   >4.56                           89.4

Tuesday                 185,000            <5                   >4.57                           90.4

Wednesday               177,500            <5                   >4.55                           92.2

Thursday                180,000            <5                   >4.56                           98.0

Friday                  160,000            <5                   >4.51                           96.2

Saturday                172,500            <5                   >4.54                           94.2

All Data                177,500            <5                   >4.55                           93.2

March 31, 2009                                                                                  Page 82 of 185
                                                     Table 4.5

                           Demonstration of spore removal using riverbank filtration (RBF)

                                                                           River             Well A   Well B

Concentration (spores/L)

Number of Samples                                                             42                58       52

Percent of Values Below Detection Limit (10 spores/L)                       none             37.9%    26.9%

Percent of Values <100 spores/L                                             none             84.5%    88.5%

Mean                                                                   8,992,890               120       97
Median                                                                   800,000                15       20

Log Removal (mean raw/mean of well)                                                             4.9      5.0
Log Removal (median raw/median of well)                                                         4.7      4.6

Log Removal Relative to River in Paired † Samples

Number of Paired Samples                                                                        26       25

Median                                                                                         4.55     4.30

Maximum                                                                                        7.26     6.48
Minimum                                                                                        1.80     2.15

March 31, 2009
                                                                         raw - old DL         raw - new DL           filt - old DL      filt - new DL



      Spore Concentration (cfu/L)






                                                    1/1                        4/1                           7/1                           9/30                         12/31

Figure 4.2                              Seasonal variation in raw and finished (“filt”) spore concentrations using historical data (old DL) and data from this study (new DL) at one of

                                        the participating utilities

March 31, 2009                                                                                                                                                     Page 84 of 185

  Cryptosporidium Log Removal


                                4                                                                                                           Huck - Calif
                                                                                                                                            Huck - Canada
                                                                                                                                            Dugan - optimized
                                3                                                                                                           Dugan - suboptimal



                                    1                   10                     100                    1,000                  10,000
                                                 Filter Effluent Spore Concentration (spore/L)

Figure 4.3                              Cryptosporidium log removal versus finished water spore concentration in pilot-scale spiking studies reported in the literature

                                Sources: Dugan et al. 2001 and Huck et al. 2002

March 31, 2009                                                                                                                                                            Page 85 of 185
                                                                                                                                      Log of Means
                                                                                                                                      Log of Medians (see Table 4.1)

                                 5                                                                                                    Log of Geometric Means
                                                                                                                                      Mean of Logs
                                                                                                                                      Median of Logs
             Spore Log Removal





                                          A              B               C              D               E               F

Figure 4.4                       Comparison of alternative methods for establishing numerical value of DOP credit using aerobic spore data from six facilities

March 31, 2009                                                                                                                                                   Page 86 of 185
                                                                       Delivered           Raw         median           mean w/ outlier             mean w/o outlier


        Percent of Observations Less Than Value
                                                                                                        mean filtered - w/ outlier included

                                                  70%                                               mean filtered - w/ outlier excluded

                                                                                              median filtered - w/ or w/o outlier
                                                                                                                                                                 mean raw

                                                  30%                                                                                                     median raw



                                                         1               10               100             1,000           10,000           100,000         1,000,000 10,000,000
                                                                                                Spore Concentration (spore/L)

Figure 4.5                                          Distribution of raw and finished water spore samples at facility “B” during calendar year 2001, including impact of a finished water outlier

                                                    (53,000 spores/L) on mean and median calculations (see Table 4.3)

March 31, 2009                                                                                                                                                                Page 87 of 185

    Percent of Observations Less Than Value

                                              70%                                                                                                         median


                                                                                           mean w/ outlier excluded


                                              20%                        mean w/ outlier included


                                                 0.00                1.00              2.00              3.00              4.00              5.00              6.00              7.00
                                                                                                       Log Spore Removal

Figure 4.6                                           Distribution of log spore removal calculated for each pair of raw and finished water samples at facility “B” during calendar year 2001,

                                                     compared with log removal calculated by taking log of mean and median raw and finished water concentrations (see Table 4.3 and Figure 4.7)

March 31, 2009                                                                                                                                                                Page 88 of 185
                                                                                    synthetic data           original censored data


                                          90%                       Concentration
Percent of Observations Less Than Value

                                          80%    Mean
                                                     Censored            7.68           4.9
                                          70%         Synthetic          7.17           4.9
                                                     Censored             <1           >5.4
                                          60%         Synthetic          0.23          6.0






                                            0.001               0.01                  0.1                    1                   10                  100                 1000
                                                                                Finished Water Concentration (spores/L)

Figure 4.7                                       Estimation of the distribution of censored (below DL) finished water spore data from facility “B” during 2003 using the methods of Helsel and

                                                 Cohn (1988) and Hirsch and Stedinger (1987)

March 31, 2009                                                                                                                                                              Page 89 of 185
                                                                                          2002        2001       2000   1999   1998   1997


       Percent of Observations Less Than Value








                                                        0.1                                       1                              10                   100
                                                                                     Finished Water Spore Concentration (spores/L)

Figure 4.8                                         Improvement over time in finished water spore concentration

March 31, 2009                                                                                                                               Page 90 of 185
                                                  Concentration (spores/L)



                                                                                                                           facility 1 - pilot-scale
                                                                                                                           facility 1 - full-scale
                                                                                                                           facility 2 - pilot-scale
                                                                                                                           facility 2- full-scale

    Finished                                 below detection limit

Figure 4.9      Pilot- versus full-scale spore removal in clarifiers and filters at two facilities

March 31, 2009                                                                                                                         Page 91 of 185
                                                                                                          Well A        Well B        River

     Percent of Observations Less Than Value (%)



















                                                                                                  Spore Concentration (spores/L)

Figure 4.10                                           Spore removal from river in two RBF wells
March 31, 2009                                                                                                                                                    Page 92 of 185

To top