SMMap: Soil Moisture Content Mapping/Estimation Using Synthetic Aperture Radar (SAR) Data

Soil moisture Content is an essential entity for plant growth and has a direct impact on crop yield. Traditional moisture estimation methods involve Gravimetric method, Water balance calculation method, Estimation using thermal and electrical conductance properties of soil. These traditional methods are accurate but can be time consuming and fail for estimation over large areas of land like forests and grasslands. A more suitable technique for estimating soil moisture over a large area is through use of Synthetic Aperture Radar which uses microwaves to map the images of land instead of visible light. This presence of moisture interferes with the backscattering of microwaves which in turn gives details about the moisture content. Existing methods involving SAR Data use empirical and semi-empirical methods to calculate SMC but these require properties of soil like soil roughness to be known beforehand. Therefore, this research work intends to employ a novel method to detect soil moisture content without coming into direct contact with it. The proposed model employs a Regression Convolutional Neural Network (RCNN) architecture that uses SAR data taken by Sentinel-1 C- band radar to forecast soil moisture content across agricultural areas. This method brings the advantage of estimating SMC over remote areas where knowing the properties of soil is not possible. The outcome of this project is a GUI based interface, where the user can upload the VV and VH images and get the predicted soil moisture output.