Causes and Prediction of Drought - NOAA-CIRES CDC GOALS by tyndale

VIEWS: 341 PAGES: 42

									               Causes and Prediction of Drought

                      Randall M. Dole
             NOAA-CIRES Climate Diagnostics Center

Drought in Great Plains, ca. 1935     Fires in the West, 2002
               Context: Importance of Drought

                         North Platte river, May 22, 2002
              Mean flow: 1310 cu. ft/sec Observed flow: 0 cu. ft/sec.
Droughts are not simply climate phenomena; they have profound societal,
economic, and environmental consequences.

Traditional federal/state response to drought has been reactive. We are
moving toward a more proactive approach - “National Drought
Preparedness Act of 2003”. Implications for NOAA wx/climate services --
                     Drought characteristics
a. Definition(s)

No unique definition. National Drought Policy Commission:
“ a persistent and abnormal moisture deficiency having adverse
impacts on vegetation, animals, and people”.

b. Types

Meteorological - rainfall deficit (better: P - E ( supply-demand))
Agricultural - topsoil moisture deficit; crop impacts.
Hydrological - surface or sub-surface water supply shortage.

Typically, meteorological            agricultural          hydrological.
Similar sequence for recovery.

For more info, on defintions, see National Drought Mitigation Center website: .
c. Time scales


Droughts span a broad range of time scales, from short-term “flash droughts”
that may have significant agricultural impacts, to multi-year or even decadal
droughts (1930s, 1950s, etc.) Paleoclimate evidence suggests that in the last 1000
years parts of the U.S. have experienced “mega-droughts” that persisted for
d. Drought indices

Numerous drought measures - all have strengths and
shortcomings. Some of the most common measures:

• Percent of normal precipitation (problem: non-normal dist.)

• Standardized precipitation index, or SPI (only considers P)

• Crop Moisture Index, CMI - simple water balance, top layer

• Palmer Drought Severity Index (PDSI) - responds slowly

• Deciles or quintiles - lowest quintile “much below normal”, need
fairly long, stable climate record.
     Examples - 2002 Case:
Standardized Precipitation Index
           (June 2002)
                  SPI local time history
              (created from WRCC website - 08/02)
                  Standardized Precipitation Index
                       Northern New Mexico


          Current 1      2        3        4     5   6
                      Years before present
       Crop Moisture Index (CMI) - 2002 Case
                   (from CPC - 24 August 2002)

The CMI is most useful for short-term monitoring (e.g., for ag.)
       Palmer Drought Severity Index - 2002 Case
                       (from CPC - 24 August 2002)

The PDSI is more useful for monitoring long-term, hydrological drought.
               PDSI- area coverage of severe drought

Percent area under severe or worse drought in western US and for Contiguous US.
Black curve based on 5-month running mean of monthly PDSI values. Red curve
shows 5-year running mean values.
         National Drought Monitor -2002 case
             represents synthesis of inputs

Most direct impacts on water supply (and demand), agriculture, fire
risk. But manifold indirect impacts as well, e.g., on recreation, energy
production, water quality, air quality, ecosystems, endangered species.
Current Crop Moisture Index
    (for period ending 7/26/2003)
Current Palmer Drought Severity Index
        (for period ending 7/26/2003)
Current Drought Monitor
      Drought Monitoring - take home points
• There is no unique definition of drought, nor is there a “best”
drought index - all have strengths and limitations.

• Consider impacts - the “human dimension”.

• Keep in mind the types of drought, and their lag relationships.

• Be wary of calling a premature end to the drought; hydrological
impacts may persist well after precipitation has returned to near

• Key factors to monitor in drought include severity, longevity,
spatial pattern and scale. Impacts will vary regionally and
depending on time of year.

•There are a number of excellent web resources available.
      Drought monitoring - some key resources
• Climate Prediction Center -

• NDMC -

• Climate Diagnostics Center -

• Western Region Climate Center -
               Drought Causes

You are asked, “ What is the cause of this drought?”

 And your answer is …
 “An unusually persistent upper-level ridge over the
General contributing factors are: anomalous subsidence, changes in
horizontal moisture transports, and shifts in the storm tracks.

Persistent upper-level ridges are often identified as the proximate cause
of drought conditions. Subsidence occurs downshear of ridge axis.
Suppresses precipitation in several ways:

   • Adiabatic warming inhibits large-scale condensation
   • Mid-tropospheric warming produces static stabilization
   • Low-level divergence inhibits moisture convergence, is frontolytical.

Even relatively weak subsidence can strongly suppress precipitation.
     Role of Anomalous Moisture transports:
                 1988 Drought

850 mb Mean Moisture Flux                  850 mb Anomalous Flux
      (April - June)                         (April-June, 1998)

              (from Lyon and Dole, 1995)
Next question:

What is the cause of this unusual flow pattern?
An important diagnostic clue: spatial pattern

The 1998-2002 drought was part of a larger global pattern.
During the same time - persistent tropical SST anomalies
            Pacific sea surface temperature on the equator, 1998-2002

 Unusual warmth
 of the “warm pool                                            Multi-year La Niña
                                                                 since 1998
       La Niña Effects on U. S. Precipitation
                 La Niña Composite (Oct.-May)
                     (created off of CDC web site)

La Niña’s effects on precipitation are most robust in SW and SE.
U.S. Oct-May Precipitation and Temp Anomalies
      Averaged for the Period 1998-2002

   Temperature                  Precipitation


Idealized experiments to test sensitivity to tropical SSTs
                  (Hoerling and Kumar, 2003)

   Simulating the Joint Impacts of Warm Indian Ocean and
    La Niña U.S. Oct-May Temperature And Precipitation

    Temperature                        Precipitation


     Observed and AGCM anomalies - specified SSTs
                        (Hoerling and Kumar, 2003)

 Observed Temperature and
 Precipitation anomalies
 (June 1998 - May 2002)

Model-simulated Temperature
and Precipitation Anomalies
given observed SSTs over this
                  Land Surface effects
• Why consider?

   • Like SSTs, LS has a “memory” beyond synoptic scales
   • After SSTs, it is most likely source for seasonal climate

   • Influence on T
   • Influence on P
          Drought changes evapotranspiration rates (ET)
                  Example: Southern Plains, Summer 1980 drought
                              (Lyon and Dole, 1995)

ET decreases during drought. Estimated anomalous heating rates: +1-2 C/ day.
Increases likelihood of heat waves. High T’s increase drought impacts (demand
                     LS effects on precipitation

Studies suggest a link between precipitation and anomalous LS
conditions, especially for major droughts and floods, such as 1980,
1988 droughts, 1993 floods.

LS conditions do not initiate droughts. They may perpetuate drought
conditions, increase the likelihood of drought recurrence (midwest),
and certainly increase drought impacts.

Main LS effects on p are likely to be in warm season.

Mechanisms are elusive: local moisture recycling, non-local effects.

Non-local effects: T’ induces PV’: flow, moisture transport,
convergence,stability change; changes in the elevated mixed layer.
       Effects of remote and local land surface processes
                   likely vary during droughts
 Studies of various droughts, and mechanistic experiments suggest that the role of
remote and land surface processes varies during the course of drought evolution.

                 (From Hong and Kalnay, 2002, for 1998 drought)
                            LS effects: Summary
• LS processes directly impact weather and climate through the surface heat and
moisture budgets. ET is the key connecting variable, and is strongly modulated
by soil moisture. At present ET is poorly observed; is estimated from models.

•The strongest and most direct LS impacts are on T, through changes in the
surface energy budget. Impacts on P are weaker, and may be due to local
moisture recycling or more subtle non-local effects.

• LS climate influences are most evident in warm season, when dynamics are
relatively weak. LS processes are likely increase the probability of summer heat
waves and may increase the duration of droughts.

• Regions that are characterized by large soil moisture variance, high ET rates,
and a dominance of convective precipitation are most likely to be sensitive to LS
processes (e.g., much of the southern and central U.S. in summer).

• There is increasing evidence that deep soil moisture may be significant in
maintaining multi-year droughts over the Great Plains (e.g., in the 1930s).
                      Other factors

• Random component - “droughts happen”.

• Forcing from other ocean regions (extratropics, Atlantic)

• Other large scale modes of variability (AO?)

• Solar variations
      The “Climate - Weather” Connection
Understanding the links between “climate” and “weather”
(precipitation) variability is vital to identifying the causes of

• To understand droughts, need to understand dominant
regional and seasonal precipitation mechanisms (synoptic-scale,
convective, orographic --)

• As droughts evolve, they may have feedback effects on T and
precip., both locally and non-locally, that effect the weather:

       - T: higher maxs., larger diurnal cycles -
       - P: “In times of drought, all signs of rain fail.”
       May see higher convective cloud bases, other effects.Non-
       local effects; e.g., on dryline, convective cap locations?
       Possible development of biases in MOS wx. products?
           Take home points - Drought Causes

• To understand droughts, it is vital to know the processes that
produce precipitation and how they are influenced by climate
variability. This will vary regionally and seasonally.

• Time scales and spatial patterns provide important clues on
drought causes.

• Several factors are likely contribute to severe and sustained
droughts, such as tropical SST forcing, land surface processes, etc.
Major factors contributing to the 1998-2002 drought were the
persistent La Niña conditions and a record warm Indian Ocean.
      Causes of drought - key web resources
• Climate Prediction Center -
  -- See expert assessments, discussions, long-lead briefings, etc.

• Climate Diagnostics Center -

 -- Applications of diagnostic tools for interpretations.
                             Drought Prediction

For operational purposes, the drought prediction problem is to
forecast the probability distribution of some quantitative drought
measure or index over a given region and time period.

Relationship to other operational products:

• Very short term - apply weekly outlooks and hazards assessments.

• CMI responds relatively rapidly - weekly to monthly forecasts relevant.

• PDSI responds slowly - monthly to seasonal and longer - seasonal forecasts.

• Consider T forecasts as well as P forecasts.
Key point: Climate forecasts are always probabilistic.
           Model-derived Seasonal Precipitation probabilities
                           for New Mexico

La Niña

El Niño

 The model results illustrate how “wet” La Niña conditions or “dry”
 El Niño conditions in New Mexico are both possible, but unlikely.
Empirical estimates of changes in risks of
    seasonal precipitation extremes
       (constructed from CDC website:

 El Niño                                La Niña
                 Drought Outlook
The current drought outlook (Climate Prediction Center):
         Drought prediction - take home points
• Climate forecasts are intrinsically probability forecasts.

• Beyond a few weeks, the major source for predictive skill is
related to changes in the distribution of tropical heating,
particularly over the Pacific and Indian Oceans. Most models
do not simulate this well.

• Users are interested in weather/climate information and
predictions across a broad range of time scales. For up to a few
weeks out, use model forecasts (consider ensembles, not just
control) plus CPC’s hazard assessment product; CMI (ag.
impacts) responds on these time scales.

•For longer-term conditions, use Drought Outlook and monthly
and seasonal forecasts. Look at IRI and CDC’s experimental
forecast pages to compare similarities and differences among
seasonal forecast models.
      Drought prediction - key resources
• Climate Prediction Center -
Hazards assessment, monthly and seasonal forecasts, drought outlooks,
experimental prediction of PDSI, etc.

• International Research Institute for Climate Prediction (IRI) -

• Climate Diagnostics Center -

Experimental forecast products week two to multi-season, model-
intercomparisons, ENSO-extreme event risks, etc.
The End

To top