Machine Learning-Based Improved WCM Technique for Soil Moisture Retrieval with Synergy of Sentinel-1 and Landsat 8 Data

Soil moisture is a crucial land-based parameter that governs the energy and mass equilibrium in land-atmosphere interactions. Unfortunately, the conventional measurement methods do not facilitate the effortless monitoring of soil moisture’s spatial and temporal variability at a regional scale. Satellite remote sensing data presents a viable solution to address the limitations and propose an alternative approach to soil moisture mapping. However, some primary challenges are the coarse spatial resolution of available soil moisture products, vegetation, and surface roughness influences. This study will implement two soil moisture estimation methods to overcome these limitations. The Water Cloud Model (WCM) is a vegetation contribution model widely used in an inversion scheme for soil moisture and crop biophysical parameters estimation over large-scale vegetated areas. However, due to the lack of a standard technique for determining these parameters, WCM must be calibrated to produce accurate estimates of crop parameters and soil moisture. Furthermore, different machine-learning models will be employed in the subsequent methodology to estimate soil moisture using Sentinel-1 SAR and Landsat-8 remote sensing data. Notably, the efficacy of these methods will be evaluated in two different sites (Iran and Canada).

Faculty Supervisor:

Saeid Homayouni

Student:

Partner:

Shahid Chamran University of Ahvaz

Discipline:

Life Sciences

Sector:

Education

University:

Université du Québec : Institut national de la recherche scientifique

Program: