A Comparative Analysis of Rainfall Prediction Model Based on GPS and BDS Signals

Rainfall prediction plays an important role in guiding agricultural development and natural disaster prediction. However, the prediction of rainfall with many influencing factors is extremely complicated, which makes high-precision prediction very difficult. This work verify the feasibility of using GPS and BDS signals to predict rainfall, and compare their accuracy.This study proposes a neural network model based on GNSS signals to predict rainfall.The paper uses the precipitable water vapor data retrieved by Global Positioning System (GPS) and Beidou positioning system (BDS), combining with neural network algorithm, to predict the actual rainfall. The experiment proves that the assumption is feasible and experimental results show that the mean absolute error of the GPS-based prediction and BDS-based prediction are 0.2-0.25(mm) and 0.15-0.2(mm), respectively. Neural network prediction model performs well.The both results are in line with meteorological requirements, while the latter accuracy is higher.