modeling and applications OF swot satellite data C. Lion1, K.M. Andreadis2, R. Fjørtoft3,F. Lyard4, N. Pourthie3, J.-F. Crétaux11LEGOS/CNES, 	2Ohio State University/JPL3CNES, 	4LEGOS/CNRS
SWOT mission1NASA and CNES, launch in 2019970km orbit, 78°inclination, 22 days repeatKaRIN: InSAR Ka bandWide swath altimeterOcean: “Low resolution” meso-scale and submeso-scalephenomena (10km and greater)Hydrology: “High resolution”surface area above (250m)² rivers above 100m970 km
2Preparing the mission for hydrologyModelisation and simulation for technical use2. SAR amplitude image:    Rhone river, France    CNES/ Altamira information simulator1. Radar cross section    CNES/ CAP Gemini simulator
GoalsNeed for a simulator for scientific users (hydrology)“Fast”: 3 months  3minEasy to use: no need for heavy preparation of input dataPortableRelatively realistic errorsTargets: deltas, rivers, lakes…Output: water elevation3Simulator output: water heightThe Amazon river, Brazil
Simulator principleBased on works of:S. Biancamaria and M. Durand: swath calculation, principleV. Enjolras: residual error calculation4
Simulator principleBased on works of:S. Biancamaria and M. Durand: swath calculation, principleV. Enjolras: residual error calculation5
Simulator principleBased on works of:S. Biancamaria and M. Durand: swath calculation, principleV. Enjolras: residual error calculation6
Residual height errors7Taken into accountRollBaseline variationThermal noiseGeometric decorrelationBAQ noiseSatellite positionNot taken into account yetTroposphereLayoverShadowProcessing (classification…)….
Residual height errors: RollRoll8BaiRr1r2Hh
Residual height errorsBaseline9E_bBiRr1r2Hh
Residual height errorsCoherence lossg = gSNR + gSQRN + ggN number of looks10BiRr1r2Hh
Simulator principleBased on works of:S. Biancamaria and M. Durand: swath calculation, principleV. Enjolras: residual error calculation11
Simulator principleBased on works of:S. Biancamaria and M. Durand: swath calculation, principleV. Enjolras: residual error calculation12m
Simulator principleBased on works of:S. Biancamaria and M. Durand: swath calculation, principleV. Enjolras: residual error calculation13
Simulation: Ohio River143 months modelizationcourtesy: K. Andreadis40.540.54040LatitudeLatitude39.539.5393938.538.5275276277278279275276277278279LongitudeLongitudeInput: Model LisFLOODReference water height (m)Output: Water height observed	 by SWOT (m)
Assimilation methodology15Assimilating SWOT observations in a identical twin synthetic experimentOhio River study domain (only main stem)LISFLOOD hydraulic modelEnsemble Kalman filterErrors introduced to boundary inflows, channel width, depth and roughnessObservation errors from a Gaussian distribution N(0,5cm)courtesy: K. Andreadis
16Assimilation resultsWater surface elevation along the river channel at two SWOT overpass times208 Hours280 HoursInformation is not always propagated down/up streamSmall ensemble size could partly be the reason  courtesy: K. Andreadis
ConclusionsSimulation of SWOT data with more representative errorsThe simulator is more user friendly: output format as input format, GUI, can be used with several modelsCan be used for assimilations studies (estimate indirect valuables)Need to improve the simulator: layover, decorrelation due to vegetation, troposphere …17
Thank for your attention

igarss2011_lion.pptx

  • 1.
    modeling and applicationsOF swot satellite data C. Lion1, K.M. Andreadis2, R. Fjørtoft3,F. Lyard4, N. Pourthie3, J.-F. Crétaux11LEGOS/CNES, 2Ohio State University/JPL3CNES, 4LEGOS/CNRS
  • 2.
    SWOT mission1NASA andCNES, launch in 2019970km orbit, 78°inclination, 22 days repeatKaRIN: InSAR Ka bandWide swath altimeterOcean: “Low resolution” meso-scale and submeso-scalephenomena (10km and greater)Hydrology: “High resolution”surface area above (250m)² rivers above 100m970 km
  • 3.
    2Preparing the missionfor hydrologyModelisation and simulation for technical use2. SAR amplitude image: Rhone river, France CNES/ Altamira information simulator1. Radar cross section CNES/ CAP Gemini simulator
  • 4.
    GoalsNeed for asimulator for scientific users (hydrology)“Fast”: 3 months  3minEasy to use: no need for heavy preparation of input dataPortableRelatively realistic errorsTargets: deltas, rivers, lakes…Output: water elevation3Simulator output: water heightThe Amazon river, Brazil
  • 5.
    Simulator principleBased onworks of:S. Biancamaria and M. Durand: swath calculation, principleV. Enjolras: residual error calculation4
  • 6.
    Simulator principleBased onworks of:S. Biancamaria and M. Durand: swath calculation, principleV. Enjolras: residual error calculation5
  • 7.
    Simulator principleBased onworks of:S. Biancamaria and M. Durand: swath calculation, principleV. Enjolras: residual error calculation6
  • 8.
    Residual height errors7Takeninto accountRollBaseline variationThermal noiseGeometric decorrelationBAQ noiseSatellite positionNot taken into account yetTroposphereLayoverShadowProcessing (classification…)….
  • 9.
    Residual height errors:RollRoll8BaiRr1r2Hh
  • 10.
  • 11.
    Residual height errorsCoherencelossg = gSNR + gSQRN + ggN number of looks10BiRr1r2Hh
  • 12.
    Simulator principleBased onworks of:S. Biancamaria and M. Durand: swath calculation, principleV. Enjolras: residual error calculation11
  • 13.
    Simulator principleBased onworks of:S. Biancamaria and M. Durand: swath calculation, principleV. Enjolras: residual error calculation12m
  • 14.
    Simulator principleBased onworks of:S. Biancamaria and M. Durand: swath calculation, principleV. Enjolras: residual error calculation13
  • 15.
    Simulation: Ohio River143months modelizationcourtesy: K. Andreadis40.540.54040LatitudeLatitude39.539.5393938.538.5275276277278279275276277278279LongitudeLongitudeInput: Model LisFLOODReference water height (m)Output: Water height observed by SWOT (m)
  • 16.
    Assimilation methodology15Assimilating SWOTobservations in a identical twin synthetic experimentOhio River study domain (only main stem)LISFLOOD hydraulic modelEnsemble Kalman filterErrors introduced to boundary inflows, channel width, depth and roughnessObservation errors from a Gaussian distribution N(0,5cm)courtesy: K. Andreadis
  • 17.
    16Assimilation resultsWater surfaceelevation along the river channel at two SWOT overpass times208 Hours280 HoursInformation is not always propagated down/up streamSmall ensemble size could partly be the reason courtesy: K. Andreadis
  • 18.
    ConclusionsSimulation of SWOTdata with more representative errorsThe simulator is more user friendly: output format as input format, GUI, can be used with several modelsCan be used for assimilations studies (estimate indirect valuables)Need to improve the simulator: layover, decorrelation due to vegetation, troposphere …17
  • 19.

Editor's Notes

  • #17 Results similar using other standard deviations for the observation error distribution