validation of aircraft and satellite remote sensing of ... · charles laymon, william crosson,...

National Space Science and Technology Center, Huntsville, AL

Validation of Aircraft and Satellite Remote Sensing of Brightness Temperatures and Derived Soil

Moisture using a Hydrologic/Radiobrightness Model

Validation of Aircraft and Satellite Remote Sensing of Brightness Temperatures and Derived Soil

Moisture using a Hydrologic/Radiobrightness Model

Charles Laymon, William Crosson, Ashutosh LimayeNational Space Science and Technology CenterHuntsville, Alabama, USA

IGARSS, Toulouse, France July 21-25, 2003


Soil Moisture: Technology DevelopmentSoil Moisture: Technology Development

Ground-basedRadiometers

PBMR

ESTAR

2DSTAR

Scale: 1 mPlot-scale experiments—verification of theoryScale: 1 mPlot-scale experiments—verification of theory

Scale: 100 mExploration of spatial mappingScale: 100 mExploration of spatial mapping

Scale: 10 kmIntegrating spatio-temporal mappingwith hydrologic research

Scale: 10 kmIntegrating spatio-temporal mappingwith hydrologic research

Scale: 100 kmMultifrequency systemsresearch

Scale: 100 kmMultifrequency systemsresearch

Beltsville, MD; Davis, CA

PSR

PALS

Soil

Moi

stur

e Se

nsin

g Te

chno

logy

SGP ’99, SMEX ’02 ,’03

Little Washita, SGP ’97

Walnut Gulch, AZ

1970 1980 1990 2000


Future Soil Moisture MissionsFuture Soil Moisture Missions

SMOS HYDROS• Launch 2005• L-band multiangular interferometry• Coverage: global• Spatial resolution: 50 km• Revisit frequency: <3 days

• Launch 2007• L-band active/passive system• Coverage: global• Spatial resolution: 3, 10 km-radar; 40 km-radiometer• Revisit frequency: 2-3 days


Advanced Microwave Scanning Radiometer for Advanced Microwave Scanning Radiometer for the Earth Observing System (AMSRthe Earth Observing System (AMSR--E)E)

Sensor Specifications:AMSR-E

• Sun-synchronous• Altitude: 803 km—AMSR, 705 km—AMSR-E • Equatorial crossing at 1030 LST for AMSR and 1330 LST for AMSR-E• Earth incidence angle of 55°• TB resampled to 25 km for soil moisture product

Orbit:

Aqua

• 12 channel, 6 frequency conically scanning passive microwave radiometer

• Built by the Japanese Space Agency, NASDA

Launch:• AMSR-E launched on Aqua on May 4, 2002• AMSR launched on the Advanced Earth Observing Satellite (ADEOS-II) on Dec. 14, 2002


ObjectivesObjectives

Objective:

Provide “best estimates” of footprint-scale mean brightness temperatures and EASE-grid-scale soil moisture with associated confidence limits.

Approach:

Utilize a coupled hydrologic/radiobrightness model to estimate brightness temperatures with data assimilation from insitu observations and airborne radiometers.

Validation is performed using data from regional experiments.

RadiobrightnessRadiobrightnessModelModel

HydrologicHydrologicModelModel

TB and soil moisture ensembles

Evaluatescaling errors

Best estimateof mean, variance,confidence limits

on TB

Model-scaleTB, SM

AMSR-scaleTB ,SM

Aggregation


Soil Moisture Experiments in 2002 Soil Moisture Experiments in 2002 (SMEX02)(SMEX02)

Location: Near Ames, IowaTime: 24 June – 13 July 2002

Regional Area:• ~5000 km2

• 48 ground sampling sites for measuring gravimetric soil moisture (daily, PM)

Walnut Creek Watershed Area:• ~ 400 km2

• 31 ground sampling sites for measuring gravimetric soil moisture, surface and soil temperatures (daily, AM), and vegetation properties (~weekly)• Surface energy flux stations, lidar, and radiosonde measurements

Iowa

Walnut CreekWatershed

AMSR-E data resampledto EASE-Grid cells

C-band data from aircraft at 800 m resolution for regional domain

L- and S-band data from aircraft at 200 m resolution for watershed domain


WatershedWatershed--scale Sampling Sitesscale Sampling Sites

Soybeans (40% of area)Corn (50% of area)


Polarimetric Scanning Radiometer (PSR)Polarimetric Scanning Radiometer (PSR)

NASA P3-BB Orion Aircraft On-board Data System

PSR Instrument

Frequency (GHz) Beamwidth5.82-6.15 10°6.32-6.65 10°6.75-7.10 10°7.15-7.50 10°


Passive and Active L and S Band Microwave Passive and Active L and S Band Microwave Instruments (PALS)Instruments (PALS)

PALS Instrument SystemNSF C-130

Parameter Radiometer RadarFrequency 1.41 and 2.69 GHz 1.26 and 3.25 GHz

Polarization V and H VV, VH, HH

Sensitivity 0.2 K 0.2 dB

Incidence Angle 45 deg. 45 deg.

Spatial Resolution ~400 m ~400 m


Measurement of Surface Soil MoistureMeasurement of Surface Soil MoistureScoops

0-6 cm profiles from sliced corescollected daily each AM at 4 points. 0

1

2

3

4

5

6

0 5 10 15 20 25 30Gravimetric Water Content

Dept

h (c

m)

Site 24 - Corn

Soil CoresImpedance Probe


Automated Monitoring StationsAutomated Monitoring Stations

TCAV

WCRsSTPsPit A

Pit B

Enclosure

2

5

10

15

20

30

WCR, STP

WCR, STP

WCR, STP

WCR, STP

WCR, STP

WCR, STP

1 TCAV2 TCAV3 TCAV4 TCAV

Depth(cm)

Sensor Depth(cm)

Sensor


Hydrologic ModelHydrologic Model

Top ofcanopy

Bare soil energy fluxes

Surface runoff Groundheat flux

Sensible LatentShortwave Longwave Precipitation

Interceptionby canopy

Canopy energy fluxes

Radiative Fluxes

Upper zone

Root zone

Bottom zone

Diffusion/drainage Heat exchange

SoilLayers

Wind

ThroughfallSensible Latent

Infiltration

Sub-surfacelateral flow


Microwave Radiobrightness ModelMicrowave Radiobrightness Model

• Based on the forward coherent wave radiative transfer model of Njoku and Kong (1977).

• Determines microwave brightness temperatures at given frequencies based on soil moisture and temperature profiles

• Soil moisture and temperature profiles are supplied by the hydrologic model or in situ observations

• Includes parameterizations for effects of surface roughness andvegetation

• Using Dobson dielectric mixing model in this study


Hydrology and Radiobrightness ModelingHydrology and Radiobrightness Modeling

• Initialize soil moisture profile based on antecedent precipitation on day 161 (10 June)

• Run SHEELS at hourly time step, forcing with NWS North Central River Forecast Center Multi-sensor Precipitation Estimates (MPE)

• Other meteorological forcing obtained from USDA Soil Climate Analysis Network (SCAN) site at Ames, Iowa

• SHEELS soil layer configuration:Zone Thickness (cm) No. layersUpper 6 6Root 94 9Bottom 50 2


Methodology: WatershedMethodology: Watershed--scale Areascale Area

FieldMeasurements

FieldMeasurements

MeteorologicalData

MeteorologicalData

HydrologicModeling

HydrologicModeling

Veg. Optical DepthSurface RoughnessVeg. Optical DepthSurface Roughness

PALS TBPALS TB

OptimalDeconvolution

Scheme


Scheme

MultifrequencyTuning

PSR TBPSR TB


Scheme


Scheme

L, S-bandTB

L, S-bandTB

C-bandTB

C-bandTB

RadiativeTransferModeling


Tune withField Mean

GSM

Forcing


Methodology: RegionalMethodology: Regional--scale Areascale Area



FieldMeasurements

FieldMeasurements

MeteorologicalData

MeteorologicalData

HydrologicModeling

HydrologicModeling

AMSR-E TBAMSR-E TB

Level 2AVersion X1Level 2A

Version X1

C, X BandTB

C, X BandTB

PSR TBPSR TB

Aggregateto 25 km

Aggregateto 25 km

C-BandTB

C-BandTB Compare

C-bandValues

C, X Band TBC, X Band TB

Aggregateto 25 km

Aggregateto 25 km

RetrieveSoil Moisture




Compare

Compare

Veg. Optical DepthSurface RoughnessVeg. Optical DepthSurface Roughness

Forcing

C, X Band TBC, X Band TB


Observed vs. Modeled Soil Moisture (0Observed vs. Modeled Soil Moisture (0--6 cm)6 cm)WatershedWatershed--scale Areascale Area

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

176 178 180 182 184 186 188 190 192 194

2002 Day of Year

ScoopsTheta ProbesHydrologic Model

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

176 178 180 182 184 186 188 190 192 194

2002 Day of Year

ScoopsTheta ProbesHydrologic Model

Gra

vim

etric

Wat

er C

onte

nt


Optimal Deconvolution of Remotely Sensed Optimal Deconvolution of Remotely Sensed Brightness TemperatureBrightness Temperature

RadiometerHorn

13° half-power beam width

Sensor response functiontranslated as gausian weight

power function

Sensor “posting”

Sensor footprint

~350 m

~100 m spacing between observations (“postings”)


Spatial Interpolation vs. Optimal DeconvolutionSpatial Interpolation vs. Optimal Deconvolution

Kriging

July 2Dry Case

July 7Wet Case

Optimal Deconvolution

July 2Dry Case

July 7Wet Case


Brightness Temperatures Derived byBrightness Temperatures Derived bySpatial Interpolation vs. Optimal DeconvolutionSpatial Interpolation vs. Optimal Deconvolution


Comparison of Methods Along a TransectComparison of Methods Along a Transect


Observed PALS TB vs. Modeled TB (1.4 GHz)Observed PALS TB vs. Modeled TB (1.4 GHz)

Mod

eled

L-b

and

TB

July 7: Wet Case

July 2: Dry Case


Polarimetric Scanning Radiometer (PSR)Polarimetric Scanning Radiometer (PSR)

July 9 July 11

440000 460000 480000

Easting (m)

4620000

4640000

4660000

4680000

4700000

4720000

Nor

thin

g (m

)

440000 460000 480000

Easting (m)

4620000

4640000

4660000

4680000

4700000

4720000

250 260 270 280 290 300

Brightness Tenmperature (Degrees K)

7.32 GHz H7.32 GHz H--PolPol

Images Provided by the USDA ARS Hydrology and Remote Sensing Lab


Radio Frequency Interference in AMSRRadio Frequency Interference in AMSR--E E at 6.9 GHzat 6.9 GHz

TB @ 6.9 GHz H-pol Descending

3 approaches for addressing RFI:• Do nothing and apply mask over affected areas• Develop/apply a RFI suppression algorithm• Develop soil moisture retrieval algorithm exclusive of 6.9 GHz information


RFI Suppression in AMSRRFI Suppression in AMSR--E TB (6.9 GHZ)E TB (6.9 GHZ)

Fremont

Cedar Rapids

Omaha

Lincoln

Peoria

Des Moines

Madison

Sioux Falls 1

2

3

4

5

6

7

8


AMSRAMSR--E TB vs. Modeled TB (6.9 GHz)E TB vs. Modeled TB (6.9 GHz)

• Mean TB for the 8-cell SMEX02 regional area• Results for horizontally-polarized 6.9 GHz frequency• Data are for ~ 1330 local time• Coincident modeled surface temperatures are shown for comparison


Summary and Future ResearchSummary and Future Research

• A modeling/data assimilation system is in place to generate ‘best estimates’ of microwave brightness temperatures and near-surface soil moisture with which to validate AMSR-E data products.

• Uncertainties in these estimates will be determined using an ensemble simulation approach.

• The hydrologic model has been tuned such that soil moisture estimates agree well with observations at the watershed scale (400 km2).

• An “optimal deconvolution” method has been developed to extract land cover based estimates of TB.

• Model TB for corn is about 10 K higher than for soybean under dry conditions and about 20 K higher under wet conditions.

• We are awaiting release of the PSR data to complete the multifrequency tuning of the radiative transfer model at the watershed scale.

• We will then run the coupled model over the regional domain in ensemble mode to generate products for AMSR validation.

• Validation of the RFI-suppression algorithm will be ongoing.• Validation activities will be repeated to some extent with data sets

generated during SMEX03.

validation of aircraft and satellite remote sensing of ... · charles laymon, william crosson,...

Documents