An open source platform to estimate Soil Moisture using Machine Learning Methods based on Eo-learn library

Abstract

Soil moisture (SM), is an important hydrologic variable that controls the interactions between land surface and atmosphere. It leads to the quantification of the quantity of water in the soil. Earth observation provides satellite data with high spatial and temporal resolution that is a key component for the estimated SM. Several Machine learning (ML) methods are used to estimate SM. In this paper, we present a platform open source to estimate SM based on Eo-learn library. The platform is developed in a complete sequence from the download of the sentinel-1A and sentinel-2A images, preprocessing, feature extraction and application of the ML models for the generation of the targeted SM data based on the Eo-learn architecture. In this work, we apply this platform to estimate SM in southern Tunisia, using the annual satellite images Sentinel-1A and Sentinel-2A from 2016 to 2017.