YHKGEWEX May June 2007 Land Surface Flux Workshop Soil moisture retrieval from space a review? Claire Gruhier, Silvia Juglea, Patricia deRosnay and Yann.

YHK GEWEX May June 2007 Land Surface Flux Workshop

Soil moisture retrieval from

spacea review?

Claire Gruhier, Silvia Juglea, Patricia deRosnay and Yann Kerr


Atmosphere Control

(Radiation)

Surface Control

(Soil Moisture)

Surface Control

(Freeze/Thaw)

Main factors controlling the water, energy and carbon fluxes

Churkina & Running, 1998


What is Soil moisture?• Water in

– the first mm– The first centimeters– In the litter layer– The root zone (Vadose)– The water table– The total– The total including interception

• Antecedent precipitaion index• An adjusting variable allowing models to run and to give ET• A soil wetness index• A red coulour in GRACE gravity fields• Does it exist?• etc


Measuring soil moisture• Laboratory

– Oven-drying soil samples – samples analysis– chemical methods

• In situ – neutron probes – resistivity– Capacity probes

• Remote sensing– SW– SWIR– IR– passive and active microwaves

Remote sensing using microwaves opens a way towards surface soil moisture mapping


ESTIMATION DE L’HUMIDITE DU SOL PAR TELEDETECTION RADAR

Soil Moisture Remote Sensing ???

P. de Rosnay, F. Baup, E. Mougin, C. Gruhier, F. Timouk, P. Hiernaux, I. Chenerie,Y. Kerr


Probing depth?

Local calibration:


Field measurements: Soil moisture dynamics

Ebangui Mallam soil moisture vertical profile before after rain: P= 61mm DW=51mm

2.5m

surf

0 12SM %m3/m3

Agoufou SM dynamicsat different depths


Field measurements for upscaling purpose: Transect measurements -> kilometric to 50km scale information

Surface Soil Moisture (SSM)- 1km transects: every 10 m.- 50km: every 1km

Total ~60 transect measurements for 2004-2006

Relationship between: local (station) & kilometric (transects) surface soil moisture

-> kilometric calibration of SSM17 18 19 20 21 220

2

4

6

8

10

12

14

16

18

Y= 3.9471x -65.5182

r=0.97rmse= 0.8 %n= 25

Automatic SSM (ms)

Tra

nse

ct S

SM

(m3

/m3 x

10

0)


Influence of Ws & roughnessActive microwaves (radar)

-20

-15

-10

-5

0

5

10 15 20 25 30 35 40 45 50

Incidence angle

back

scat

ter

coef

f (dB

)

-20

-15

-10

-5

0

5

15 20 25 30 35 40 45 50

incidence angle

back

scat

ter

coef

f (d

B)

s =1,0.8,0.6,0.4

cm

Ws=30,25,20,

15 %

P. Waldteufel 1999


1 - En Actif: ENVISAT/ASAR

MALI (Agoufou): Baup et al 2007a et 2007b

Niger (Kori de Diantandou): (Zribi et al 2006)

0

5

10

15

20

0 5 10 15 20

Soil moisture measurements (%)

Est

imat

ed m

oist

ure

(%)

HH

VV

Rms=2.2%

Précision d’inversion: 2.8% et 2.2% sur les 2 sites

Résol: ~ 1km, 15j


- Methodological development and validation over Agoufou intensive site- Use of WS mode, HH polarization- Normalization function accounts for vegetation angular effect:

- Dry season (Nov-May)- Wet season (Jun-Oct)

Active Microwave Remote Sensing: ENVISAT/ASAR

15 20 25 30 35 40 45-19

-18

-17

-16

-15

-14

-13

-12

-11

Incidence angle (°)

Ba

cksc

att

eri

ng

Co

eff

icie

nt

(dB

)

Wet SeasonDry Season

Baup et al. RSE 2006 (submitted)


Inter-annual variations of backscatter (ERS-WSC) in the Sahelian Inter-annual variations of backscatter (ERS-WSC) in the Sahelian areaarea

Jarlan et al.


ERS-WSC / NDVI VGT Comparison in SahelERS-WSC / NDVI VGT Comparison in Sahel

Jarlan et al.


Soil Surface

Vegetation

Atmosphere

Tsky

TT

T soil

vegveg

TBskyTBveg

TBsoil

TBobs TBsky TBveg TBsoil= + +

{

Galactic

TEC

O2, IWC, T, ILWC, rain, and resp heights

Tv, , WC, structure,S

,Ts,S/C,,

SM

0 1 2 3 4 5

Canopy Water Content (kg/m2)

0

50

100

150

200

250

300

Brig

htne

ss T

empe

ratu

re (

K)

Total Brightness

Soil

Vegetation

Sky


TB sensitivity

Kerr, 1997


Influence of WV on radiometric sensitivity to Ws

• In order to penetrate vegetation cover, lowest possible frequencies should be selected.

• F < 1GHz : large Faraday effects

• Then the protected frequency is :

1.41 GHz !

Kerr and Jackson, 1995


0 5 10 15 20 25 30 35

Soil Moisture (% by Weight)

220

230

240

250

260

270

280

Bri

gh

tne

ss T

em

pe

ratu

re (

K)

Skylab S-194 Data1.4 GHz Microwave Radiometer

Independent SamplesModel Relationship

Eagleman and Lin (1976)

Jackson 2001


Wigneron et al., 95

Retrievals


Passive Microwave Remote SensingInversion of AMSR Brightness temperature

Statistical inversion based on multiple configuration observation of the surface, provided by AMSR polarization H and V.

SM* = a0 + a1 PR +a2 (TBv-TBh) ; SM* retrieved Surface Soil Moisture PR = (Tbv-TBh)/(TBv+TBh) ; Polarization ratio from AMSR BT measurementsCost function based on minimization of RMSE

Validation of correlation, RMSE, Efficiency, Mean Bias

On going: investigate further statistical inversion (multi freq C & X bands) and addresse semi-empirical regression based on C-band modeling

Range -1 to 1 (Best agreement)


Passive Microwave Remote Sensing: AMSR

AMSR on Gourma Supersite : NASA products overestimate SSM

Specific SSM inversion is developed based in statistical regression of bi-polarized measurements.

Validation preliminary results:

RMSE(%m3/m3) Correlation Efficiency MeanBias (%m3/m3) 2006 2005 AMSR-NASA 4.32 0.54 -0.8 3.3 5.04 0.45 -0.37 3.25AMSR-retrieved 2.77 0.5 0.25 0.01 3.8 0.46 0.21 -0.03 -> Statistical retrieval slightly improves the accuray of AMSR SSM compared to NASA products – Further investigations required to take advantage ofmultispectral (C & X bands)


SMOSREX and Sud-OuestFrance, Toulouse

Comparison between AMSR-E and surface soil moisture over 3 networks for 2005


GoREx (NAFE)Autralia, Goulburn River catchment



AMMA – Gourma regionMali, Agoufou



Passive Microwave Remote Sensing: AMSR

Aug 2nd decade

Aug 1st decade

July 3nd decade

AMSR-E Land 3 NASA products (Njoku et al 2003): Soil Moisture :

-> 1) Evaluation with ground soil moisture network and upscaling measurementsRadiometric measurements (Brightness temperatures):

-> 2) Development of statistical inversion over the Gourma super site


EVOLUTION OF MEAN SOIL MOISTURE FOR JANUARY (SMMR + AMSR)

SMMR AMSR

YEAR

FOR JULY

SM X10


EVOLUTION OF SOIL MOISTURE FOR JULY SMMR (regression AMSR)


Issues• Topography• Mixed pixels• Sub pixel heterogeneity• Saturations (water, forests, snow?)• Root Zone Soil moisture,• Freezing• litter• A word about forests• ...


RadiativeTransfer.Model of TB

InteractiveSVATMODEL

Données Climatiques(Rayont, Ta, ua, qa, ..)

VegetationLAIBiomass

W2_init

TB,

wS, TS

w2, Fluxes

Soil / VegetationParameters

Données Climatiques(Rayont, Ta, ua, qa, ..)

Remote Sensingmeasurements

Climatic Data(Rg, Ta, ua, qa, ...)

Interactive Vegetationparameters:

-Mes. Conductance GM

-Biomass/LAI ratio: BsL-effective life Duration: DE

OpticalDepth =K. LAI

COUPLING INTERACTIVE VEGETATION MODEL (ISBA-AGS) and RT MODELS

Wigneron et al., 2001

YHK GEWEX May June 2007 Land Surface Flux WorkshopWigneron et al 2000


Coupling and disaggregation Scheme

Wg mean, W mean

t

SVATSIMPLE

LE, Rn,H,Gpercolation

Infiltration

Subsurface flow

Saturation excess Runoff

TOPMODEL

Wg mean, W meant + dt

{ Wg i }, { W i }t + dt Soil proprieties

+

DTM

Pellenq et al., 2001


Spatialised Disaggregation Results

OBSERVED SIMULATED TOPOGRAPHY

SIMULATED TOPOGRAPHY+ SOIL DEPTH

DOY 610

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.00 0.10 0.20 0.30 0.40 0.50OBSERVED WATER CONTENT (v/v)

SIM

UL

AT

ED

SO

IL W

AT

ER

CO

NT

EN

T (

v/v)

R2=0.8950.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.50

0.00 0.10 0.20 0.30 0.40 0.50OBSERVED WATER CONTENT (v/v)

SIM

UL

AT

ED

SO

IL W

AT

ER

CO

NT

EN

T (

v/v)

R2=0.88

Correlation:

Pellenq et al., 2001


Simulated TB 1km

Simulated TB 1km55km footprint55km

NIGER: Meso scale TB simulations: ISBA + C-MEB (1km²)C-band Microwave Emission of the Biosphere (Pellarin et al., 2007, in prep)

Agrégation/désagrégation

Simulated TB25km-reggrided

AMSR-E TB Level 325km product

Résol ~ 25km/1j -> désagrégation à 1km/1j


Use of Altimetry

• Principle– Measure water heights wit use of alytimeters, – Gives amount of water

• Issues– Requires width and shape of bottom– Validity of water masks– Small bodies wrt resolution


Water fractions and ESRI


Use of gravimetry• Principle

– Measures changes of gravimetric field– Gives changes in water stocks

• Issues– A signal is there but – Never validated– Strange results not yet explained


Seasonal changes in gravity as seen by GRACE (truncated at n = 10) and predicted from a mean hydrology model (from Andersen et al. 2005a), unit of the amplitude is μGal (= 10-8 m s-2) unit of the phase is days wrt January first.


Standard deviation (in μGal) of the annual changes from 5 different hydrology models (from Andersen et al. 2005a).


Hydrological load (in mm of water equivalent) from the GLDAS model (left) and inducedgravity changes (in μGal) (center) and ground vertical displacement (in mm) (right) for March andSeptember 2003.


Differences for March and September 2003 of the gravity field (in m.s-2) betweenhydrological models and GRACE observations; upper left: raw LadWorld model corresponding at atruncation of degree n = 180 (1° x 1°); upper right same but truncated at n = 15 and lower centerGRACE observations truncated at n = 15.


Other issues

• Validation

• Sea variations– Validation experiments

• Antarctica

• …..


Conclusions?

• Gravimetry– Needs validation

• Active SAR– Relative values at best.– See scat

• Active Scat– Very complementary– Freeze thaw

• Altimetry – Complementary

• Passive low freq– Needs aux data– Still some issues

Why should P=ET+R+S?


Thank you …..Any questions ? http//www.cesbio.ups-tlse.fr/us/indexsmos.html

YHKGEWEX May June 2007 Land Surface Flux Workshop Soil moisture retrieval from space a review? Claire Gruhier, Silvia Juglea, Patricia deRosnay and Yann.

Documents

soil moisture retrieval

surface soil moisture

baup et

measurements kilometric

saheljarlan et

zribi et

sahelian areajarlan

m3m3agoufou sm