Docstoc

Parsiani-group-presentation cal-val symposium UW-July-2005

Document Sample
Parsiani-group-presentation cal-val symposium UW-July-2005 Powered By Docstoc
					 Calibration/Validation Efforts
           at UPRM


           Hamed Parsiani,
Electrical & Computer Engineering Department
    University of Puerto Rico at Mayagüez
            parsiani@ece.uprm.edu
                 Content
• Validation of cloud top height retrieval by
  MODIS and MISR instruments
• Calibration of Radar Remote Sensing as
  Applied to Soil Moisture and Vegetation
  Health Determination
• SEAWIFS validation in costal waters of
  western Puerto Rico
• Validation of Hydro-Estimator and the
  Tropical Rainfall Prediction
                      Validation of cloud top height retrieval by
                            MODIS and MISR instruments
                          Ramon Vasquez, Hamed Parsiani (Ana Picon)

Cloud top heights can be good indicators of the presence of
different types of clouds over a region.
 The information about clouds tops provides an input to some
climate models that can predict total water content.
The Caribbean data of the MODIS were obtained from the EOS Data
Gateway (EDG).
Available lidar instrumentation does not provide sufficient
information about cloud profiles. However, Cross-comparisons of
MODIS and MISR instruments can retrieve cloud top heights.




                                                                  Cross-comparison between
     MODIS cloud top heights and MISR stereo heights.             MODIS and MISR.
                          RESEARCH RESULTS
Cloud Top Height Retrieval from the Earth Observing System (EOS) Sensors

• variations between MODIS and
MISR cloud top heights may
indicate the retrieval of two different
cloud heights over the same area.
• MISR retrieval performance for
high clouds is twice the MODIS
retrieval performance.
• MISR and MODIS cloud values
coincide in less than 1% of the total
observed area and the cloud height
value is 14km.
• Results show the ability of MODIS
to detect low clouds at tropical
regions.                                       Temporal analysis that shows the variation
• MISR is a better instrument to               of MODIS cloud top heights over San Juan,
measure high clouds. MODIS                     Puerto Rico
retrieval methods can identify
thicker clouds which are low clouds        latitude 13.1 N , 35.6 S, longitude -66.2 W, -60.6 E
and MISR retrieval methods can
                                                              Retrieval Percentage Rate(%)
identify thinner clouds which are
                                      Sensor      High clouds          Mid clouds          Low clouds
high clouds.
                                         MODIS         13.98             16.81              12.25
                                         MISR          26.89              8.94               9.05
           Calibration of Radar Remote Sensing as Applied
        to Soil Moisture and Vegetation Health Determination
     Hamed Parsiani (Mairim Torres, Enrico Mattei, Allen Lizarraga)
•   The Material Characteristics in Frequency Domain (MCFD) algorithm
    calculates the MCFD for each GPR image which is used as a signature to
    determine soil moisture, soil type, and vegetation index. The usage of
    properly trained Neural Network acts as a calibrator for the GPR in soil
    moisture, or soil type determination.
•   Vegetation Health is obtained by calibrating the power of MCFD, using the
    linear relationship between the NDVI obtained by spectroradiometer and the
    MCFD power.
•   The range for calibration and its accuracy for the vegetation health have
    been determined.
•   The basic accuracy in both soil characteristics and vegetation information
    depend on the reception of images with quality wavelets. An algorithm is
    developed which permit Automatic Quality Wavelet Extraction (AQWE).
    Currently a 1.5 GHz antenna has been used for this research.
       GPR Produced Image

GPR
operation at
1.5 GHz




                             GPR image


  Example:      Air/Sand
                surface
  Subsurface    reflection

  Image
  produced by
  GPR
Vegetation Health
     Index
                       Moisture Determination and validation database, based on Ground Penetrating Radar Measurements

                                   MCFD-NN Moisture Vs. Theta Probe Moisture                                                   MCFD Power vs. NDVI

                       25.00                                                                                      0.6
                                   Trained Moisture                                                                             y = -1.1214x + 1.0792
                                   Tested Moisture
                                                                         y = 0.9931x + 0.0911
                                                                                                                  0.5
                                                                                                                                      R2 = 0.9861
                                                                              R2 = 0.9991
MCFD-NN Moisture (%)




                       20.00
                                   Linear (Tested
                                   )Moisture          y = 0.8442x + 2.3279                                        0.4




                                                                                                     MCFD power
                                                           R2 = 0.9246
                       15.00
                                                                                                                  0.3


                       10.00                                                                                      0.2


                                                                                                                  0.1
                        5.00


                                                                                                                   0
                                                                                                                    0.5        0.6       0.7          0.8   0.9    1
                        0.00
                               0            5                10              15          20     25                                             NDVI
                                                      Theta Probe Moisture (%)




 Advanced Land Observing Satellite
 computer representation. Which                                                                                           Ground Penetrating Radar @ 2 GHz
 includes PRISM(stereo mapping), AVNIR
 (infrared radiometer), and PALSAR                                                                                        High speed soil moisture determination
 (L-Band aperture radar)
                 SEAWIFS VALIDATION IN COASTAL WATERS
                       OF WESTERN PUERTO RICO
                      Fernando Gilbes (Patrick Reyes)

•   Mayagüez Bay is a semi-enclosed bay in the west coast of Puerto Rico that suffers
    spatial and temporal variations in phytoplankton pigments and suspended sediments
    due to seasonal discharge of local rivers.
•   New methods and instruments have been used as part of NOAA CREST project,
    allowing a good understanding of the processes affecting the signal detected by
    remote sensors.
•   A large bio-optical data set has been collected during several cruises in Mayagüez
    Bay. Remote Sensing Reflectance, Chlorophyll-a, Suspended Sediments, and
    absorption of Colored Dissolved Organic Matter (CDOM) were measured spatially
    and temporally. These values were used to evaluate SeaWiFS OC-2 and OC-4 bio-
    optical algorithms in the region.
•   Remote sensed Chlorophyll-a concentrations were compared against in situ
    Chlorophyll-a concentrations. The results show that these algorithms overestimate
    the actual Chlorophyll-a.
•    It is clearly demonstrated that the major sources of this error is the variability of
    CDOM and total suspended sediments. The main working hypothesis establishes a
    possible relationship between CDOM and the clays in those sediments.
•   The analyses of SeaWiFS images also verify that its spatial resolution is not
    appropriate for these coastal waters. The available data demonstrate that improved
    algorithms and different remote sensing techniques are necessary for this coastal
    region.
•   We plan to continue these efforts to validate and calibrate ocean color sensors in
    Mayagüez Bay, like MODIS and AVIRIS. We aim to improve the remote sensing
    techniques for a better estimation of water quality parameters in coastal waters,
    specifically Chlorophyll-a, CDOM absorption, and suspended sediments.
                                                               Bio-optical Properties and Remote Sensing of Mayagüez Bay

VALIDATION OF SEAWIFS ALGORITHMS IN
  MAYAGÜEZ BAY FOR CHLOROPHYLL-A
                          12.0


                                       y = 3.0325x + 0.1654
                                           R2 = 0.6871         OC2
                          10.0                                 OC4
                                       y = 4.1792x + 0.1248
                                           R2 = 0.6895
 Estimated Chl-a (ug/l)




                           8.0




                           6.0




                           4.0




                           2.0




                           0.0
                                 0.0                     0.5         1.0               1.5   2.0   2.5
                                                                                                         ABSORPTION COEFFICIENT OF CDOM
                                                                     Measured Chl-a (ug/l)
                                                                                                                      1 .0

                                                                                                                                                                       St at ion   1
                                                                                                                      0 .9                                             St at ion   4
                                                                                                                                                                       St at ion   5
                                                                                                                                                                       St at ion   7
                                                                                                                      0 .8                                             St at ion   9
                                                                                                                                                                       St at ion   11
                                                                                                                      0 .7                                             St at ion   13
                                                                                                                                                                       St at ion   15
                                                                                                                                                                       St at ion   17
                                                                                                                      0 .6                                             St at ion   19
                                                                                                                                                                       St at ion   21
                                                                                                         ag ( m-1 )



                                                                                                                                                                       St at ion   23
                                                                                                                      0 .5


                                                                                                                      0 .4


                                                                                                                      0 .3


                                                                                                                      0 .2


                                                                                                                      0 .1


                                                                                                                      0 .0
                                                                                                                         350   400   450   500         550     600   650                700
                                                                                                                                           Wavelengt h ( nm)
   Validation of Hydro-Estimator and the Tropical Rainfall Prediction
  Nazario Ramirez & Ramon Vasquez (Beatriz Cruz)

                                                   (a)
 This is the first time that the Hydro-
Estimator (HE) algorithm is validated over a
tropical region.
 The USGS monitors, in Puerto Rico, 120           (b)
rain-gauges & records rainfall every 15
minutes.
 Estimation of precipitation was
generated by the same spatial and                  a)    From rain gauge (24 Hrs) observed data
temporal distribution using the HE                 b)    Hydro estimator (HE)
algorithm.                                               Comparison between H-E vs rain gauges
                                                                  ( Nov. 11-13, 2003.)

Preliminary results:


 HE algorithm underestimates heavy
precipitation
 A correlation coefficient of 0.6 is
observed between estimated and observed
rainfalls.

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:8
posted:3/4/2010
language:English
pages:11