A Low-cost Soil Moisture Monitoring Method by Using Walabot and Machine Learning Algorithms

Abstract

Soil moisture plays an important role in agricultural processes, which has a significant effect on crop evapotranspiration, the exchange of water, and energy fluxes. Recently, soil moisture can be measured by remote sensing or proximate sensing techniques, such as thermal, optical, and microwave measurements. However, there are limitations to the applications of these methods, such as low spatial resolution, limited surface penetration, and vegetation. In this study, it proposed a new low-cost soil moisture monitoring method by using a Walabot sensor and machine learning algorithms. Walabot is a pocket-sized device cutting-edge technology for Radio Frequency tridimensional sensing. Unlike the remote sensing tools such as unmanned aerial vehicles (UAVs) limited by cloud cover or payload capability, the Walabot can be used flexibly in the field and provide data information more promptly and accurately than UAVs or satellite. By putting different moisture levels of soil on the Walabot, the Walabot can collect radio frequency reflectance data from different levels of soil moisture. Then, machine learning algorithms, such as principal component analysis (PCA), linear discriminant analysis (LDA), have been applied for data processing. Results showed that Walabot has a state-of-art performance in estimating soil moisture.