Strategies for Unlocking Knowledge Management in Microsoft 365 in the Copilot...
MO3.T03_3399_GRUHIER.pdf
1. SMOS SOIL MOISTURE
PRODUCT EVALUATION
OVER VARIOUS SITES
Claire Gruhier, Arnaud Mialon, Silvia Juglea,
Ahmad Albitar, Simone Bircher,
Thierry Pellarin, Yann Kerr
CESBIO, Toulouse, France
DTU Space, Copenhagen, Denmark
LTHE, Grenoble, France
IGARSS 2011-8283 | HS6.2 | Monday 25 July 2011 | Vancouver, Canada
2. Introduction
Soil Moisture and Ocean Salinity
Passive microwaves sensor in L-band
Launched the 2th November 2009
Global soil moisture product every 3 days
In the context of the CAL/VAL of SMOS mission
Evaluation of SMOS soil moisture product
allows us to improve algorithm and finally product
CAL/VAL over various sites around the world
A lot of sites was chosen to conduct CAL/VAL activities
In this study : 6 sites
Denmark HOBE Hydrologycal Observatory
South of France SMOSREX Surface Monitoring Of Soil Reservoir EXperiment
East of Spain VAS Valencia Anchor Station
West Africa AMMA African Monsoon Multidisciplinary Analyses
East of Australia AACES Australian Airborne Cal/Val Experiment for SMOS
USA SCAN Soil Climate Analysis Network
3. Denmark HOBE
Location
Skjern River Catchment
Measurements
Airborne campaign EMIRAD (L-band, HV)
4 flights & ground sampling, 26 april to 9 may 2010
30 stations within SMOS ”pixel”, since Dec 2009
Environmental conditions
Climate : Temperate-maritime
Land cover : Crop 78%, Forest 14%, Heath 6%
Soil : sandy
S. Bircher et al, "A soil
moisture network for SMOS
validation in the Skjern River
Catchment, Western
Denmark", in prep.
S. Bircher,et al, "SMOS
Validation by means of an
airborne campaign in the
Skjern River Catchment,
Western Denmark",
submitted to TGRSS, SMOS
Special Issue (under revision).
Simone Bircher, DTU
4. Denmark HOBE
Airborne campaign: TB comparison over spatial scales
Model ground (L-MEB) vs EMIRAD EMIRAD avg vs SMOS L1C HV
2x2 km patch scale 44x44 km SMOS ”pixel” scale
300
May 2, 2010 04:15 UTC
GROUND TB (K)
H_0 H_40
280
260
TB (K)
240
270 220
V_0 V_40
GROUND TB
200
245
Crop 180 EMIRAD V EMIRAD H
Heath x SMOS V x SMOS H
(K)
Forest 160
220 245 270 0 10 20 30 40 50 60 70
EMIRAD TB [K] EMIRAD TB [K] Ө°
Able to reproduce EMIRAD Good accordance between RMSE (K)
measurements by means of modeled EMIRAD and SMOS TBs 0° 40°
TBs from ground data on all 4 Other campaign days too H 7.74 9.33
campaign days, for all patch types RFI-prone for comparison.. V 8.30 5.58
Simone Bircher, DTU
5. Denmark HOBE
Comparison SM values at pixel scale
> Retrieved L2SM product and network 0.30
In situ values (m3/m3)
soil moisture show the same trend
> Retrieved SMOS SM exhibits higher 0.20
amplitudes and a distinct negative bias
compared to the ground data
0.10
R = 0.7
RMSE = 0.096
Bias = 0.086
0.0 0.10 0.20 0.30
L2SM (m3/m3)
0.35
SM (m3/m3)
0.0
Jan 10 Mar May Jul Sep Nov Jan 11
Simone Bircher, DTU
6. South of France SMOSREX
Location
Near Toulouse (Mauzac, ONERA site) 43°23 N ; 1°17 E
Measurements
LEWIS L-band radiometer (HV), since 2003
Ө = 20°, 30°, 40°, 50 °, 60°
Soil moisture/temperature network, 2 stations, since
2003
Environmental conditions
Climate : Temperate
Land cover : Bare soil
and Grassland
Soil : 16% clay, 36% sand
Arnaud Mialon, CESBIO
7. South of France SMOSREX
Comparison SM values
SMOS vs In situ
> The same trend can be observed January to May 2011
between L2SM product and the station
Bare soil Grassland
> L2SM values are closer to in situ R 0.64 0.58
values on bare soil in term of absolute RMSE 0.076 0.208
values and dynamic Bias -0.051 -0.199
0.5 0.5
L2SM (m3/m3)
SM (m3/m3)
0.0
Jan 10 Fev Mar Apr May 0.0 0.5
In situ
(m3/m3)
Arnaud Mialon, CESBIO
8. East of Spain VAS
Location
Utiel-Requena Plateau
Valencia, Spain, 39°34’15’’N, 1°17’18’’W
Measurements
Soil moisture/temperature network,
6 stations, since 2007
Environmental conditions
Climate : semiarid and dry-sub-humid
Land cover : vineyard crops, olive and almond trees
surrounded by pine and Mediterranean forests.
Soil : sand
125x125 km²
50x50 km²
10 km
10 km
SMOS pixel
50 km
size
125 km
50 km
10x10 km²
125 km
control area
Silvia Juglea, CNES/CESBIO
9. East of Spain VAS
Comparison of TB simulated with L1C product
> Soil moisture is spatialised
with SURFEX model
TB are modeled with L-MEB
> The temporal variations of
TBhv SMOS is in good
agreement with the TBmodel
on VAS, but TB from SMOS are
too high
> The parameterization of the
radiative transfert model
must be reconsidered to take
into account all the
characteritics of the VAS area
TB SMOS vs TB VAS
H V
R2 0.51 0.42
RMSE 16.24 20.82
Bias 0.02 14.72
Silvia Juglea, CNES/CESBIO
10. East of Spain VAS
Comparison SM values at pixel scale
> L2SM values are lower than those modeled by SURFEX (validated with insitu
values). This underestimation can be explained by too high SMOS TB (could be
explained by RFI)
SM in-situ vs L2SM
July to November
2010
18h 6h
DES ASC
R2 0.59 0.54
RMSE 0.109 0.085
Bias -0.076 -0.039
Silvia Juglea, CNES/CESBIO
11. West Africa AMMA
Location
Niger and Benin
Measurements
Soil moisture/temperature network, 6 stations
Environmental conditions
Climate : Sahelian / Sudanese
Land cover : tiger bush / woody savanna and tropical
forest
Soil : sand
Gruhier C., T. Pellarin, P. de Rosnay, Y Kerr, « SMOS soil moisture
.
product evaluation over West-Africa at local and regional scale »
Claire Gruhier, CESBIO
12. West Africa AMMA
Comparison SM values over Niger site
> Comparison to weighted averaged
of in-situ values according to their
distance from the DGG selected
Height = DQX
Width = STD of in-situ measurements
> Same trend
> RMSE less than 0.04 m3/m3
(Gruhier et al, submitted)
Claire Gruhier, CESBIO
13. West Africa AMMA
Comparison SM values over Benin site
> Same trend
> RMSE higer
than over Niger
because of
variability during
rainy season
due to forest
land cover
(Gruhier et al, submitted)
Claire Gruhier, CESBIO
14. East of Australia
Location
AACES
Australia, Murrumbidgee Catchment
Measurements
Airborne campaign AACES (L-band, HV)
January/Feb. 2010 and September 2010
Permanent soil moisture stations
Environmental conditions
Climate : Semi-arid
Land cover : crops and grassland
Soil : sand
> see Chris Rugider's talk
Arnaud Mialon, CESBIO
15. East of Australia AACES
Comparison SM values
> SMOS L2SM closer from in-situ values
than initial values (ECMWF)
> In general, SMOS L2SM less than In
situ soil moisture values. Except after
rain events
> SMOS sensitive to 0-3 cm top surface
layer, whereas in situ measurements are
0-5 or 0-8 cm surface layer
Arnaud Mialon, CESBIO
16. East of Australia AACES
Comparison SM values
> SMOS L2SM closer from in-situ values
than initial values (ECMWF)
> In general, SMOS L2SM less than In
situ soil moisture values. Except after
rain events
> SMOS sensitive to 0-3 cm top surface
layer, whereas in situ measurements are
0-5 or 0-8 cm surface layer
Arnaud Mialon, CESBIO
17. USA SCAN
Location
United States of America
Measurements
235 permanent soil moisture stations
Environmental conditions
Climate : Various
Land cover : Various (Forest, crops, grassland...)
Soil : Various
Ahmad Albitar, CESBIO
18. USA SCAN
Comparison SM values
> Same trend can be
observed between both soil
moisture series
> Tau values are questionable
> SMOS captures the dry
downs
Nb values 230
R 0.77
RMSE 0.065
Bias 0.04
Ahmad Albitar, CESBIO
19. USA SCAN
Complete comparison Correlation Correlation
all sites (235) nominal sites
(98)
> Nodes with more than 90%
of bare soil and low
vegetation
Sites = 98/235
> Good fit across a variety of
local sites but complete
analysis for all sites
decreases performances
Bias RMSE
> SMOS soil moisture is dryer nominal sites nominal cases
than site data (98) (98)
reasons : penetration
depth /measured depth, data
quality, spatial averaging
Ahmad Albitar, CESBIO
20. Conclusion
> L2SM product based on V4 algorithm clearly improved the accuracy of
8
retrieved values and more values are also retrieved (not shown)
> Soil moisture variations provided by L2SM V4 product are consistent with in-
situ measurements in term of correlation and RMSE
The rainy events are perfectly reproduced
The L-band sensitivity allow us to monitor drying out period
> SMOS L2SM product generaly underestimate ground measurements,
which can be explained by the depth:
In-situ values are recorded at 5cm of depth
the penetration depth of SMOS is ~0-3cm
by scaling effect and representativity:
Local measurement / spatial integrated values
And RFI in some areas
> Reduction of RFI source is on the way
> Mironov/Dobson models are currently being evaluated to computed the
emissivity
> Forest modeling is being investigated to improve algorithm for high vegetation
optical depth
> … and more
> SMOS after 14 months in operation is alearldy giving good reasults but there
is still room for improvements... and we are working on it !
21. SMOS SOIL
MOISTURE PRODUCT
EVALUATION OVER
VARIOUS SITES
Claire Gruhier, Arnaud Mialon, Silvia
Juglea,
Ahmad Albitar, Simone Bircher,
Thierry Pellarin, Yann Kerr
CESBIO, Toulouse, France
DTU Space, Copenhagen, Denmark
LTHE, Grenoble, France
IGARSS 2011-8283 | HS6.2 | Monday 25 July 2011 | Vancouver, Canada