Analysis of CYGNSS Soil Moisture Retrieval Based on ANN over a Selected Region in Ethiopia

Soil moisture (SM) plays a vital role in agriculture, ecosystem functioning, water conservation, weather predictions and climate models. High spatial and temporal frequency data of soil moisture is crucial for agricultural and other important applications. Recent advancements have brought attention to the possibility of using GNSS reflectometry (GNSS-R) for applications on land such as snow sensing, soil moisture retrieval, sea surface monitoring and other applications in addition to positioning, navigation, and timing applications of GNSS. Cyclone Global Navigation Satellite System (CYGNSS) is designed to improve hurricane forecasting by studying the interaction between the ocean and the atmosphere within tropical cyclones. However recent studies show the opportunity of this system for high spatio-temporal soil moisture retrieval. This study presents a machine learning-based approach to get SM at a selected region in Ethiopia using CYGNSS data and analysis of the result. Artificial Neural Network (ANN) model is developed and used to predict soil moisture. The Soil Moisture Active Passive (SMAP) global soil moisture data have been used as reference data in the ML algorithm. The proposed approach has achieved a good correlation between predicted values of soil moisture and reference values from SMAP.