Assimilation of TerraSAR-X data into a snowpack model

Xuan-Vu Phan (1) , Michel Gay (2, 1) , Laurent Ferro-Famil (3) , Yves Durand (4) , Marie Dumont (4)

1 GIPSA-SIGMAPHY - GIPSA - Signal Images Physique
2 GIPSA-Services - GIPSA-Services
3 IETR - Institut d'Électronique et des Technologies du numéRique
4 CNRM - Centre national de recherches météorologiques

OBJECTIVES • Implementation of a multilayer snowpack electromagnetic backscat-tering model (EBM), based on Dense Media Transfer Radiative (DMRT), at high frequency bands (X-band and above). • Application of 3D-VAR data assimilation to constrain the snow meta-morphism model Crocus using SAR image data and the EBM. ELECTROMAGNETIC BACKSCATTERING MODEL • The simulated snowpack backscattering consists of 3 main backscattering mechanisms: 0 sim = 0 as + 0 vol + 0 g • 0 as and 0 g are calcu-lated using Integral Equa-tion Model (IEM). • 0 vol -Total volume backscattering of all snowpack layers, derived from DMRT equations. 3D-VAR DATA ASSIMILATION Meteorological model SAFRAN SAR acquisition SURFEX/Crocus Electromagnetic Backscattering Model (EBM) H(x) = σ sim Minimization of cost function J Assimilated variables x Adjoint model íµí»»H t 3D-Var data assimilation Guess Variables x g Optimised yes no y = σ obs Meteorological variables 3D-VAR assimilation adjusts the guess parameters to reduce the dis-crepancy between sim and T SX , according to the error statistics of modeling and observations. In order to constrain the initial guess data using the observations, one needs to minimize the cost function J: J = (x x g) t B 1 (x x g) + (y H(x)) t R 1 (y H(x)) where : • x g and x -the initial guess and analysed snowpack parameters, contains the values of density and grain size of each snow layer. • y and H(x) -SAR calibrated backscattering (T SX) and simulated backscattering coefficient (sim). • B and R -the error covariance matrices of Crocus and observations. This work has been funded by GlaRiskAlp, a French-Italian project (2010-2013) on glacial hazards in the Western Alps and MeteoFrance, the French national meteorological service. TerraSAR-X data was provided by German Aerospace Center (DLR). Crocus snow evolution . data was provided by CEN, MeteoFrance. In-situ mea-surements were carried out by IETR (University of Rennes 1), Gipsa-lab (Grenoble INP) and MeteoFrance.

Domaines: Traitement du signal et de l'image [eess.SP]

Xuan-Vu Phan, Michel Gay, Laurent Ferro-Famil, Yves Durand, Marie Dumont. Assimilation of TerraSAR-X data into a snowpack model. IGARSS 2014 - IEEE International Geoscience and Remote Sensing Symposium, Jul 2014, Quebec, Canada. pp.3998 - 4001, 2014, ⟨10.1109/IGARSS.2014.6947362⟩. ⟨hal-01118455⟩