Study on Rapid Inversion of Soil Water Content from Ground-Penetrating Radar Data Based on Deep Learning

Ground-penetrating radar (GPR) is an efficient and nondestructive geophysical method with great potential for detecting soil water content at the farmland scale. However, a key challenge in detection obtaining rapidly real-time. In recent years, deep learning methods have become more widespread earth sciences, making it possible to use them inversion from GPR data. this paper, we propose neural network framework GPRSW based on of end-to-end that directly inverts volumetric (VSWC) through single-channel Synthetic experiments show accurately identifies different VSWC boundaries model time depth. The predicted fit well within 40 ns, maximum error after ns less than 0.10 cm3 × cm?3. To validate our method, conducted measurements experimental field Academy Agricultural Sciences Gongzhuling City, Jilin Province applied measurements. results values match samples are consistent overall trend TDR probe samples, difference not exceeding 0.03 Therefore, study shows has be obtain data farmland.