The SMOS Soil Moisture Retrieval Algorithm

YHK 2006 Pan Gewex Meeting October 9-13 ESRIN It

SMOS Soil moisture

retrieval Algorithm

Y. H. Kerr1, …and the SMOS team


Outline

• SMOS mission overview in a nutshell

• Soil moisture retrievals– Level 2– Level 3 – Level 4

• Results, issues and next steps

• Conclusion


Ocean salinity rationale• Thermohaline overturning circulation.

How can climate variations induce changes in the global ocean circulation?• Air-sea freshwater budget.

How are global precipitation, evaporation, and the cycling of water changing?• Tropical ocean and climate feedback

Science Objectives for SMOS: Salinity

Lagerloef et al., 2001


• Role of Soil moisture in surface atmosphere interactions:

storage of water (surface and root zone), water uptake by vegetation (root zone), fluxes at the interface (evaporation), influence on run-off

• Implies relevance forWeather ForecastsClimatic studiesWater resources crop managementForecast of extreme events

• Climate change predictions and rain event forecasts requires SST and SM

Science Objectives for SMOS: Soil Moisture


Data Assimilation ExperimentsM. Drusch ECMWF

1. CTRL OI (Optimum Interpolation) based on screen level analyses

for the top three model soil layers.

2. OL (Open Loop) without any soil moisture analysis.

3. NUDGE (Nudging) experiment using the TMI Pathfinder soil

moisture product.Common features:

- Full atmospheric 4DVar analysis using ~ 106 observations / 6h

(reflecting the operational configuration).

- Model version CY29R1.

- T511 spectral resolution, 60 vertical levels.

- ‘Early delivery’ set up with 10-day forecasts from 00 and 12 UTC.

- Study period from 1 June to 31 July 2002.


Soil moisture increments (CTRL OI)

80°S80°S

70°S 70°S

60°S60°S

50°S 50°S

40°S40°S

30°S 30°S

20°S20°S

10°S 10°S

0°0°

10°N 10°N

20°N20°N

30°N 30°N

40°N40°N

50°N 50°N

60°N60°N

70°N 70°N

80°N80°N

160°W

160°W 140°W

140°W 120°W

120°W 100°W

100°W 80°W

80°W 60°W

60°W 40°W

40°W 20°W

20°W 0°

0° 20°E

20°E 40°E

40°E 60°E

60°E 80°E

80°E 100°E

100°E 120°E

120°E 140°E

140°E 160°E

160°E

-250

-200

-150

-100

-50

-10

10

50

100

150

200

250

[mm]

Accumulated root zone soil moisture increments for June 2 to July 30, 2002.

Analysis increments are a sizeable part of the terrestrial water budget.


Validation of soil moisturearea averages for Oklahoma (72 stations)

surface soil moisture

model forecast (OI)

observationsmodel forecast (OL)

• Too quick dry downs (model problem).• Too much precip in July (model problem).• Too little water added in wet conditions

(analysis problem).• NO water removed in dry conditions

(analysis problem).

root zone soil moisture

model forecast (OI)

observationsmodel forecast (OL)

• Precipitation errors propagate to the root zone.

• Analysis constantly adds water.• The monthly trend is underestimated.

The current analysis fails to produce more accurate soil moisture estimates.

M Drusch


Validation of soil moisturearea averages for Oklahoma

surface soil moisture

• Nudging / satellite data remove water effectively and produce a realistic drydown.

• Nudging the satellite results in the mostaccurate surface soil moisture estimate.

root zone soil moisture

• The information introduced at the surface propagates to the root zone.

• The monthly trend is well reproduced using the nudging scheme.

Satellite derived soil moisture improves the soil moisture analysis and results in the most accurate estimate.

M Drusch


How to access soil moisture• Ground measurements• Networks (GEWEX)• SW, SWIR, ….• THIR….• Low frequency microwaves

– Active microwaves• Vegetation, roughness• Revisit• Sensitivity

– Passive microwaves antenna issue


Science Objectives for SMOS: The SMOS Mission

•Need for soil moisture and sea surface salinity fields•Only passive L band suitable•Real aperture systems currently not adequate (antenna size)==>Synthetic antenna

Mission specificationsSoil Moisture

multi-angulardual pol

4 % vol 3 day revisit(Vegetation 7 day)better than 50 km

TB(40,V) 1st February 87 (K)

TB(40,V) 1

st February 87 (K)

TB(40,H) 1

st February 87 (K)

0 7200

300

TB(40,H) 1st February 87 (K)

TB(40,H) 1

st February 87 (K)

TB(40,H) 1

st February 87 (K)

0 7200

300

Sea Surface Salinitymulti acquisitions

dual pol or 1st stokesbetter than 0.1 psu10 day to monthly

Grid scale (200 km)

SMOS is the second Earth Explorer opportunity mission (1st round)An ESA/CNES/CDTI projectSelected in 1999, initiated in 2000 Phase B finished, C/D Started in January 2004 for a launch in 2008A new technique (2D interferometry) to provide global measurements from space of key variables (SSS and SM) for the first time.

Pellarin et alLe Traon et al


Interferometry• angular resolution

provided by distantantennas

• correlation productss(1)*s(2) → visibilityfunctions V(D/λ)

• Inverse F.T. on V → TB(θ)

Space sampling requirement : every λ/2 value at least onetime ; hence "thinning" possibilities.

1

θ

DD '

∆θ

∆θ ≈ λ / D'

2


•Each integration time, (2.4 s) a full scene is acquired (dual or full pol)•Average resolution 43 km, global coverage•A given point of the surface is thus seen with several angles•Maximum time (equator) between two acquisitions 3 days

Principle of operations

SMOS FOV; 755 km, 3x6, 33°, 0.875λ,

P. Waldteufel, 2003


SMOSBaseline

• 18+5 elements /armconfiguration

• Mean Altitude --> 757 km +/- 500 m• 30° steer angle• 32.5° tilt angle • Elements spacing

0.875 λ• Dual or Full pol

2006 Pan Gewex Meeting October 9-13 ESRIN It

Payload Module (deployed)

Payload Module (stowed)SMOS in Rockot

CASA EADS, 2003

SMOS Instrument: MIRAS derived conceptCASA EADS (Spain)

BUS: PROTEUS Alcatel Space Industry

Launcher ROCKOTGround segment: Level 0-2 Villafranca

Level 3-4 Toulouse

Esa lead with contributions from France and Spain


Schedule• Current:

– Payload module is being assembled (CASA)– Antennas have been measured (TUD)– Ground segment at various stages depending levels– Delivery of PLM after testing early 2007– Level 1 Prototype operational end 2006– Level 2 Breadboards-prototypes Mid 2007– CATDS (Level 3 and 4) initiated end 2007

• Overall schedule– Launch 2008– 6 months commissioning phase– 2.5 + 2 years operation

• Cal val AO• Science use AO• Near Real Time• SMOSOps


Hardware


Schedule• Current:

– Payload module is being assembled (CASA)– Antennas are being measured (TUD)– Ground segment at various stages depending levels– Delivery of PLM after testing end of year– Level 1 Prototype operational end 2006– Level 2 Breadboards-prototypes Mid 2007– CATDS (Level 3 and 4) initiated end 2007

• Overall schedule– Launch 2008– 6 months commissioning phase– 2.5 + 2 years operation

• Cal val AO• Science use AO• Near Real Time• SMOSOps


HUT Demonstrator

Brightness temperature (K) measured by HUT-2D over densely forested forestareas surrounded by open field areas. Right: land use classification for the measuredarea. High brightness temperature areas correlate well with forest areas, and lowerbrightness temperature areas correlate well with open areas. The flight altitude is 1500meters resulting in a ground resolution of approximately 100 x 100 meters.Juha Kainulainen, et al. 2006


Technical Elements

Payload Data Processing CenterOperations Data Flow – Open Loop Concept

S-bandX-band

XBASX-Band Acqusition Station

TTCETTM & TC Earth Terminal

SOCCSpacecraft Operations

Control Center

PDPCPayload Data Processing

Centre (LO-L2)ESOC

CATDSCentre Aval de Traitementdes DonnéesSmos (L3-L4)

USERS

L0,L1,L2

L3,L4Reports

L2,L3,L4Requests

Pass Plans S/C & PL commandsTTCET commands

Pass plans

HKTM-PHKTM-R

TTCET Monitoring Data

Calibration Data & PL status

PL-HKTM, PL-TM, Pass plans & Orbit determination information

Villafranca Kiruna

IF

SC status & PLTM S/C & PL commands

HKTM

A. Hahne Estec

ESLExpert Support

Laboratory


spatial Data products available will be:

Level 1: brightness temperature at H and V polarisation

Level 2: daily soil moisture and ocean salinity (swath) maps at basic temporal and resolutions

Level 3: daily global soil moisture maps global salinity maps

Level 4: special products (root zone SM, SSS (Godae) fields, HYDROS / AQUARIUS / AMSR mix etc,

Other: Improved algorithms (level 1 and 2), and calibration, CalVAl

Services

All data products are produced (with quality statement) and distributed to registered users.

• All data products are archived for the duration of the mission plus 10 years.

• All data products are in a catalogue with a browse facility.

Data products


Aim and boundary conditions• Derive Soil moisture and Vegetation optical thickness

(Water content) and …depending upon data available• Characterisation of angular signature (max up to over

100 measurements per pixels ( 10) down to a very few at the end of the swath)

• Two polarisations every 3 days• Basic resolution 43 km on average (~25 to 50 km)• Higher spatial sampling 15 km typically (nodes)• Equal area projection• Blackman apodisation Weighting function/area• Many surface types per integration area /pixel

elementary surfaces


Models and mixed pixels

global B T at SM O Santennas

S ynthetic antenna d irectional ga ind i

e1 e2 e3 e4 e5 e6 e7 e8 e9

G 2

G 5

type 1 em itter ⇒ C F 1



antenna footprin t

G 7

: Aggregated fractions FM0 and FM

FM0 class Aggregated land cover FM class Complementarity A B C D E

FNO Vegetated soil + sand FNO FFO Forest FFO FWL Wetlands FWL

Open fresh water FWP Open saline water FWS

FWO Open water FEB Barren FEB FTI Total Ice fraction

Ice & permanent snow FEI Sea Ice FSI

FUL Low urban coverage

FUM Moderate urban coverage

Comple-mentary

Sum of comple-mentary fractions equals unity

FTS Strong topography FTM Moderate topography FRZ Frost FRZ

Non permanent dry snow Non permanent wet snow FSN Non permanent mixed snow

FSN

Supple-mentary

Supple-mentary fractions are super-imposed

Set of models fror each surface type



0 5010 4020 30

Teodosio Lacava – IMAA-CNR;

Mean Soil Moisture (% volume): January 2003-2005 A

Projection: EASE Grid Global; Datum: WGS-84


Teodosio Lacava – IMAA-CNR

Projection: EASE Grid Global; Datum: WGS-84

0 5010 4020 30

Mean Soil Moisture (% volume): April 2003-2005 A


Pre Processing filter SMOS views co-locate auxiliary data with DGG node compute aggregated mean cover fractions compute reference values

Iterative parameters retrievalfit the selected forward model to observations use other models for default contribution

Decision Tree select main fraction for retrieval select one model for retrieval select other models for default contributions

SMOS L1c Data products files Auxiliary files

SMOS L2 auxiliaryFixed reference Evolving reference User parameters file

TB models libraryforward models default models future models

Post Processing compute parameters posterior variances retrieval analysis and diagnostics compute modeled surface TB @ 42.5° generate flags

DGG node data iterator

Generate output data

User Data Productretrieved parameters parameters variances science & quality flags

Data Analysis Report

algorithms internal datafull flags

Do full retrieval again eliminate bad views decrease free number of parameters

Level 2:Soil moistureAlgorithm

layout


Issues: What is inside a pixel?• Decision tree set of models

– Fixed features:• Low vegetation, forested, barren, water, permanent snow/ice,

urban– Variable features

• Snow, variable water bodies/ floods, frozen ground/ vegetation

– Overlapping perturbing factors• Topography, RFI

– Issues:• Knowledge• Temporal signatures


External data


The SMOSworld….And DGGs…

DFFG @ δkm

=300 km, for φ ∈ {−87,87} & λ ∈ {0,360}

Orthographic Projection


Pure pixels?...

15oW 10oW 5oW 0o 5oE 10oE

42oN

45oN

48oN

51oN

54oN WA_DGG_exact

Colour code/ ● ocean, ● inland water body, ● rivers, ● All others, ● WA_DGG_nowater;


What is in a Pixel?!!!Working Area @ DGG ID #32002432 [φ = 44.42°, λ = −1.115°, 123 km x 123 km]

EEAP DFFG @ δkm

= 4 km, Zone #72

Albers Equal−Area 2oW 30’ 1oW 30’ 0o

40’

44oN

20’

40’

45oN

Working area contains all

the footprints and more

Nodes spatial sampling

Elementary surfaces (DFFG)


The land use map (MétéoFrance –Ecoclimap binned into generic classes)How homogeneous ?areas with prevailing nominal and forested aggregated LC fractions


In conclusion: SM retrievals can be attempted in many areas with varying

expected accuracy


Algorithm validation PlanDirect simulation Lo-

op Selection x / + N data set

Data options for DQX runs 792 TBL1Surface scene

Scene homogeneous nominal ; suggested code 166WEF none necessary

Input parameters for simulated dataSM SL 0.0 to 0.5 step 0.1 x 6TAU TL1 0, 0.01, 0.025, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8 x 12Sand, clay 0.4, 0.3Tsurf 288Tsky 3.7T0 atm, P0, WVC 288, 1013, 3ro_s, alpha, SAL, eps_pa From UPF LUTHR, NR_H, etc. from CLASS LUT for selected code

Input observation conditionsRadiometric uncertainty simulated based on L1c test scenarioX_SWATH (km) XL 0, 100, 200, 300, 350, 400, 425, 450, 475, 500, 525 x 11Incidence & rotation angles from L1c test scenarioFaraday angles from L1c test scenarioL1c flags from L1c test scenario

Additional data options for bias runs 726Radiometric uncertainty Zeronon uniform TAU Mixing 2 direct models with TAU=0 and 0.6 + 1 TBL3

non uniform SM TL2 Mixing 2 direct models with 2 SM TBD around loop value, for TAU=0 to 0.6 step 0.2 + 4 TBL4

Faraday angle TL2 Double L1_TS values; TAU=0 to 0.6 step 0.2 + 2 TBL5Rotation angle TL2 2° uniform; TAU=0 to 0.6 step 0.2 + 4 TBL6SM SL 0.0 to 0.5 step 0.1 x 6X_SWATH XL 0, 100, 200, 300, 350, 400, 425, 450, 475, 500, 525 x 11

Total 1518

Need to validate the algorithm and the products:

A) Theoretical scenes, homogeneous, all over the swath, exact data acquisition, typical noise figure, exact geometry

B) Step in SM and vegetation opacity over the whole range

C) Same analysis on realistic synthetic data sets

Global and high resolution15 days at several dates (solstices and

equinox)

D) Validation on ground data sets and then real sat data during commissioning phase

The Murrumbidgee Catchment

SM retrieval performance with free TAUSM

SM retrieval performance with constrained TAU


Cal Val• Initiated (2005)• Teams selected (2005)• Optimisation process under way• Rely on CEOP whenever possible• GEWEX soil moisture network• Possibility to add sites• Work with modellers


NAFE ’05 J Walker et al.Regional Data: 1km

796

183

Elevation

Visible


Issues• Water bodies• Varying features (Frozen, snow)• Perturbations (rain interception)• RFI• Unknown surfaces (urban, barren ….)• And things to discover….


AMSR DATA RFI at 6.8 (blue) and 10.7 (green)


AMSR DATA RFI at 6.8 (blue) and 10.7 (green)


23 January 2003LEWIS

L band for Estimating

WaterIn

Soils

High QualityGround based radiometer

1.4 GHz, H & Vsensitivity 0.1 K

main lobe 12.5 @3db, 22° beam efficiency 0.986No « visible » back lobes

ONERA/CESBIOOperational since 23/1/2003


Level 3 product• Basis : maximum possibilities: full resolution • Compatibility with other « products »• With well defined sampling on a fixed grid

– Composite over well defined periods– Typical sampling

• 1,3, 6-7, 10, 30days 1 year• 40 km, 1 °

– But updated daily in several cases– GODAE like

• And tool boxes• Common format with other missions?• From ideally level 2 but actually L1 more likely


Level 4• Mainly research• Merged products (other existing missions)• Synergisms with future missions• Inter-calibration• Merge with models (run off, dis -aggregation, assimilation,

Mercator etc…)• Stress on coastal areas, , Sea ice, Ice?, flooded areas,

Catchments, flood /drought risks flags• Fire prone areas• Sky Map• Sun activity /tec• Non exhaustive list……….

AQUARIUSAQUARIUSSea Surface Salinity Mission


Data Assimilation ProductsData Assimilation Products

Land Surface Modeled PhysicsDataData

Model IntegrationsModel Integrations

UpdateUpdate

UpdateUpdate

Global at 5 kmat model time-step

L4_5km_4DDAL4_5km_4DDAproductproduct


Atmospheric Forcing(Rg, Ra, Ta, qa, U, P) Remote sensing

Soil and vegetationparameters

LAI

SVAT model

SURFACESoil moistureTemperature

VegetationOpticalthickness

InteractiveVegetation

Radiative Transfermodel

Root-zonesoilmoisture

Plant growthparametersor biomass

Radiances

OPTIMISATION

RootRoot Zone Zone soilsoil moisturemoisturePRINCIPLE OF THE ASSIMILATION OF REMOTE PRINCIPLE OF THE ASSIMILATION OF REMOTE

SENSING DATASENSING DATA

Calvet et al.


Dis-aggregation• With use of higher resolution data (O. Merlin 2005)• Topo and vegetation+ SVAT model (J. Pellenq 2003)

Measured SM (SGP ’97)

Dis aggregated SM (O Merlin 2005)

SMOS pixel 40x40 km

AVHRR Pixels TIR

1 km

Dis-aggregationPixel to pixel comparison


Conclusion• SMOS will be the first mission to deliver global fields of soil moisture

and sea surface salinity• It is an EXPLORER Mission ==> new concept new instrument new

measurements!!• The challenge NO data exists, NO Algorithm exists: we are breaking

new grounds• Launch date 2008 - tomorrow!• Payload and Bus well underway• Ground segment started• Campaigns underway (CoSMOS NAFE and SSS, AMMA…)• Specific scientific activities (modelling, exotic targets, wiggles etc..)• Cal Val AO initiated

– ( http//www.cesbio.ups-tlse.fr/us/indexsmos.html)• Issues: NRT, RFI• SMOS Ops


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

The SMOS Soil Moisture Retrieval Algorithm

Documents