the cosmos/nafe campaign for the validation of the smos
Post on 26-Jan-2022
3 Views
Preview:
TRANSCRIPT
The CoSMOS/NAFE campaign for the validation of the SMOS retrieval algorithm
over land
1kauzar.saleh@geog.cam.ac.uk
K. Saleh, Y. Kerr, P. Richaume, J.P. Wigneron, S. Delwart, M.J. Escorihuela, P. Maisongrande, G. Boulet and the CoSMOS/NAFE team
Microrad’08-
Firenze 11-14 March 2008
2
1/
ESA CoSMOS study (2007-2008): Overview
EVALUATION OF THE
SMOS SM LEVEL 2 (L2)ALGORITHM USING
AERIAL DATA
VALIDATIONOF MICROWAVEMODEL L-MEB
‘SPECIAL ISSUES’(e.g. sun-glint over land,
interception …)ANCILLARY DATA?
3
1/
ESA CoSMOS study (2007-2008): OverviewTraining
for upcomingSMOS
Validationcampaigns
EVALUATION OF THE
SMOS SM LEVEL 2 (L2)ALGORITHM USING
AERIAL DATA
‘SPECIAL ISSUES’(e.g. sun-glint over land,
interception …)
VALIDATIONOF MICROWAVEMODEL L-MEB
SENSITIVITY TOANCILLARY DATA?
1/
CoSMOS-NAFE Experiment: Overview & objectives
6-week campaign 2 L-band sensors onboard two aircrafts (2 week overlap)Focus area: ~60 km x 60 km farming area
Collaboration between NAFE experiment (University of Melbourne), researchers from Europe, US and Australia, ESA
Goulburn River catchment
4
Focus farms area ~ 60 km x 60 km
5
CROP AREAS
VWC: 0.3-2.7 kgm-2
Merriwa Park Illogan
Pembroke Illogan
6
GRAZING AREAS
VWC: 0.1-1 kgm-2
Dales Dales
Stanley
7
2/
Instruments
a) ESA flights-
EMIRAD
EMIRAD POLARIMETRIC RADIOMETER (TUD) onboard Aero Commander 500S
8
Nadir~ 40 deg
Full-StokesIR Radiometer onboard
-30
-20
-10
0
10
20
30
-180 -120 -60 0 60 120 180
Across Flight Track Off-Nadir Angle ( deg )
Am
plitu
de (
dBi )
3L 2L 1L 1R 2R 3R
<< Left Right >>
b) NAFE flights-
PLMR
PLMR RADIOMETER (UM) onboard Diamond Eco-Dimona motor glider
9
2/
Instruments
+/-7,
+/-21.5,
+/-38.5
H & V pol
3/
L-band data processing
Tbi,j
1Antenna projected onto the surfacef(aircraft attitude,antenna orientation,no slopes)
c c cg g g gg g g gg g g g
cff
2Land-use from Landsat classif.
(e.g. crop,grass,forest)+
Ancillary surface data
5Weighting by
antenna gain &integration
4Transportation of simulated Stokes vector to antenna
level
3
Tbi,j
Simulated Tbi,jusing SMOS BBs
6Comparison to
measured Stokesand retrievals
10
3/
L-band data processing
Simulations of TBand retrievals
take into account antenna pattern
andobserved surface
11
Large aperture : 3dB area up to ~1 km (off-nadir)
13/11 14/11 15/11 17/11
20/11 21/11 22/11 23/11
27/11 29/11 03/12 09/12
EMIRAD
EMIRAD
rainfall
rainfall
EMIRAD EMIRAD
EMIRAD
EMIRAD EMIRAD EMIRAD
EMIRAD
EMIRAD
EMIRAD
4/
L-BAND measurements[K]
12
PLMR along-track02/11,09/11,16/11,23/11
4/
L-BAND measurements
13
EMIRAD
-Dry to wet soil ΔTB~40 K at nadir (for ΔSM~ 0.3 m3m-3)-Low angular variation in TB between Nadir and 40 deg ΔTBv < 5-10 K for dry soil, ΔTBv < 15-20 K for wet soil -Rather stable TB measurements across the farm for dry soil
nadir
40 deg
5/
Modelling
What do L-MEB simulations indicate?
SM=0.38 (σ=0.07) m3m-3 SM=0.22 (σ
=0.05) m3m-3
High-res area
14
PLMRAlongtrack
Wheat,Clay-loam
5/
Modelling
What do L-MEB simulations indicate?
SM=0.38 (σ=0.07) m3m-3 SM=0.22 (σ
=0.05) m3m-3
High-res area
Wheat,Clay-loam
14
PLMRAlongtrack
5/
Modelling
What do L-MEB simulations indicate?
EMIRAD and PLMR over the same wheat area on the same day - Dry soil(green, TB from Hr and τn
from PLMR)(black, same but for τn
= 0.06*LAI)
15
EMIRADAlongtrack
PLMRAlongtrack
Tsfce=294 KTsfce=304 K
5/
Modelling
What do L-MEB simulations indicate?
16
Γ= Γ
*exp(-Hr/cosNp
(θ))
HR~0.6, ploughed soil,Clay-loam, black basalts
•Literature: crops Hr=[0.1-0.7]•Sensitivity of HR to dielectric model •No clear sensitivity of HR to SM
PLMRAlongtrack
Explore next if default b’(LAI) too low for wheat?
b’=τn
/LAI
•Field VWC & b=0.08 --> τnwheat [ 0.05-0.1]
Tau from LAI
Tau from VWC
Tau free
6/SM Retrievals
CROPS Clay, HR=1 ΔSM_station ~ 0.3 m3m-3
Rain -5 d Rain -7 d Rain -10 d
Rain -0 d Rain -0 d Rain -4 d
EMIRADAlongtrack
17
6/ SM Retrievals
CROPS Clay, sparse grass- HR=1 ΔSM_station ~ 0.3 m3m-3
Rain -5 d Rain -7 d Rain -10 d
Rain -0 d Rain -0 d Rain -4 d
Rain -5 d Rain -7 d Rain -10 d
Rain -0 d Rain -0 d Rain -4 d
EMIRADAlongtrack
Silt-loam,crops , HR=1 ΔSM_station ~ 0.3 m3m-18
19
7/
Conclusions
i)
L-band processing ‘close’
to the SMOS L2 approach(aggregated TBs within the antenna pattern, SMOS BBs)
• Basis for upcoming airborne campaigns over land
ii)
The CoSMOS study is now in the algorithm validation phase
• Focus on small areas with dense sampling to characterisevegetation and soil emission, next step is moving to thefarm scale
iii)
Both PLMR and EMIRAD data produce a TB signaturethat requires large Hr
• Hr is usually fitted in experiments, how do we fit it globally?• Hr = f(land use, soil type, std)
iii)
Once roughness is fitted, changes in the retrieved and measured SM are comparable
SM, T,Hr,Tau
SM, T,Hr,Tau
SM, T,Hr,Tau
Measured
Retrieved
MWvalidation
SMvalidation
7/
Conclusions
20
4/
Modelling
What do L-MEB simulations indicate?
Mean HR=0.6, ploughed soil,70% clay, black basaltsLiterature: crops Hr=[0.1-0.7]Field WC & b=0.08 --> Hr~1, higher TB error
17
3/
L-band data processing
Zone 2. Farm scale
Cell-to-cell comparisons between surface data and L-band retrievalse.g. comparisons to SM when availablee.g. LAI and retrieved optical depth
Gridding pixel 3dB area
ScatteredSM data
1x
Figure 6b. Temperatures at Merriwa Park: Yellow: IR station (when available), Black: surface 2.5 cm, Dotted line: Flights PLMR
c) Data set overview
10
2/
Summary of acquired data-
L-BAND (EMIRAD AND PLMR ALONG-TRACK)
5/
Modelling
What do L-MEB simulations indicate?
wet
PLMR -along-track high-res area
High-res area
Grass,clay
top related