synergy of l-band and optical data for soil moisture monitoring o. merlin, j. walker and r. panciera...
TRANSCRIPT
Synergy of L-band and optical data for soil moisture monitoring
O. Merlin, J. Walker and R. Panciera
3rd NAFE workshop 17-18 sept. 2007
Objective
Use synergy optical/passive microwave
for improving
1. Accuracy (passive microwave scale)
OR
2. Spatial resolution (downscaling)
of L-band derived soil moisture retrievals
Data
• Regional area of NAFE’06
• 1km resolution PLMR data: TB
• 1km resolution MODIS (Terra/Aqua) data: Tsurf, NDVI
TB SMRetrieval algo
LAI
Illustration
Impact of vegetation on TB: multi-spectral retrieval
Sensitivity of Tsurf to SM:downscaling
TB/SM
Tsurf
Downscaling algo
SM
Synergy L-band/optical
1. SM retrieval
RT model:- TAU-OMEGA formalism Mo et al., 1982
- soil roughness (H,Q) Wang and Choudhury, 1981
- Teff = f(Tsurf,T2) Wigneron et al., 2001
- TAU = bVWC Jackson and Schmugge, 1991
Inverse model:Minimize (TBobs - TBsim)2
SMRetrieval algo
Tsurf
MODIS
TB
PLMR
LAI
MODIS
Teff = f(Tsurf,T2)
VWC = 0.5 LAI
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
0.0000 1.0000 2.0000 3.0000 4.0000
1. SM retrievalApplication to NAFE’06 regional area (Yanco)
Assumptions:veg para, roughness uniformStanding water = Bare soil with SM 100% v/v Retrieval algo
TBH Angle
SM
Tsurf LAI
1. SM retrievalComparison with ground measurements at the PLMR scale
Model parameters:Sand = 30%Clay = 30%b = 0.15OMEGA = 0.05T2 = 20degCH = 0.1
RMSE = 3.2% v/v Bias ~ 10 % v/v
70% of pixels 30% of pixels
2. SM downscaling
Downscaling algo
MODIS
SM
MODIS
SM
NDVI
Tsurf
Test a downscaling technique of ~40km SMOS like data from MODIS data
2. SM downscaling
minsoil,maxsoil,
soilmax,soilSEFTT
TT
Approach: SEF (soil evaporative fraction) as a proxy of surface soil moisture
MODIS SEF derived from triangle method
Ta
Tmax
NDVI
Tsurf
NDVImin NDVImax
Tsoil
2. SM downscaling
MODISMODISSMOS SEFSEFWfWW
Modified downscaling relationship
One difficulty: the non-linearity of SEF to SM
Generated SM (% v/v)
EF
(%
v/v
) SEF
2. SM downscaling
Modified downscaling relationship
SEF model Komatsu, 2003
arWc
WSEF
1exp1
0model
MODISMODISmodelSMOS SEFSEFWfWW
2. SM downscalingCorrelation between MODIS SEF and PLMR SM
SM sensitivity of Tsurf ~ SM sensivity of TB /10
2. SM downscalingLimitations and applicability:
Dry-end conditions (Tmax)
Uncertainty in SEF is high: need to aggregate to lower resolution
Could account for heterogeneity of soil
ConclusionsIllustrated two applications of the synergy between optical and passive microwave data
Preliminary SM product with accuracy ~4% v/v for 70% of the validation area (fitted with roughness H)
An example of downscaling technique of SMOS type data from 1km MODIS type data
Some questions: - stripes on PLMR TB images- bias in retrieved SM over 30% validation pixels (not explained by any parameter)-…