Skip to Main content Skip to Navigation

Suivi des ressources en eau par une approche combinant la télédétection multi-capteur et la modélisation phénoménologique

Abstract : This thesis aims to improve the spatio-temporal resolution of surface water fluxes at the land surface-atmosphere interface based on appropriate models that rely on readily available multi-sensor remote sensing data. This work has been set up to further develop (disaggregation, assimilation, energy balance modeling) approaches related to soil moisture monitoring in order to optimize water management over semi-arid areas. Currently, the near surface soil moisture data sets available at global scale have a spatial resolution that is too coarse for hydrological applications. Especially, the near surface soil moisture retrieved from passive microwave observations such as AMSR-E (Advanced Microwave Scanning Radiometer-EOS) and SMOS (Soil Moisture and Ocean Salinity) data have a spatial resolution of about 60 km and 40 km, respectively. In this context, the downscaling algorithm "DISaggregation based on Physical And Theoretical scale Change" (or DisPATCh) has been developed. The near surface soil moisture variability is estimate within a low resolution pixel at the targeted 1 km resolution based on an evapotranspiration model using LST (Land surface temperature) and NDVI (vegetation index) derived from MODIS (MODerate resolution Imaging Spectroradiometer) data. Within a first step, DisPATCh is applied to SMOS and AMSR-E soil moisture products over the Murrumbidgee river catchment in Southeastern Australia and is evaluated during a one-year period. It is found that the downscaling efficiency is lower in winter than during the hotter months when DisPATCh performance is optimal. However, the temporal resolution of DisPATCh data is limited by the gaps in MODIS images due to cloud cover, and by the temporal resolution of passive microwave observations (global coverage every 3 days for SMOS). The second step proposes an approach to overcome these limitations by assimilating the 1 km resolution DisPATCh data into a simple dynamic soil model forced by reanalysis meteorological data including precipitation. The original approach combines a variational scheme for root-zone soil moisture analysis and a sequential approach for the update of surface soil moisture. The performance is assessed using ground measurements of soil moisture in the Tensift-Haouz region in Morocco and the Yanco area in Australia during 2014. It is found that the downscaling/assimilation scheme is an efficient approach to estimate the dynamics of the 1 km resolution surface soil moisture at daily time scale, even when coarse scale and inaccurate meteorological data including rainfall are used. The third step presents a physically-based method to correct LST data for topographic effects in order to offer the opportunity for applying DisPATCh over mountainous areas. The approach is tested using ASTER (Advanced Spaceborne Thermal Emission Reflection Radiometer) and Landsat data over a 6 km by 6 km steep-sided area in the Moroccan Atlas. It is found that the strong correlations between LST and illumination over rugged terrain before correction are greatly reduced at ~100 m resolution after the topographic correction. Such a correction method could potentially be used as a proxy of the surface water status over mountainous terrain. This thesis opens the path for developing new remote sensing-based methods in order to retrieve water inputs -including both precipitation and irrigation- at high spatial resolution for water management.
Document type :
Complete list of metadatas
Contributor : Abes Star :  Contact
Submitted on : Tuesday, December 12, 2017 - 12:08:19 AM
Last modification on : Thursday, October 15, 2020 - 3:19:20 AM


Version validated by the jury (STAR)


  • HAL Id : tel-01473141, version 2



Yoann Malbéteau. Suivi des ressources en eau par une approche combinant la télédétection multi-capteur et la modélisation phénoménologique. Hydrologie. Université Paul Sabatier - Toulouse III, 2016. Français. ⟨NNT : 2016TOU30193⟩. ⟨tel-01473141v2⟩



Record views


Files downloads