基于PSO-RF的GNSS-IR土壤湿度反演方法研究

Research on GNSS-IR Soil Moisture Inversion Method Based on PSO-RF

基于全球导航卫星系统干涉测量(Global Navigation Satellite System Interferometry and Reflectometry,GNSS-IR)的土壤湿度监测弥补了传统测量方法的不足,是近年来遥感领域研究的热点。针对以往研究多采用全球定位系统(Global Positioning System,GPS)的观测量数据来估计土壤湿度,且反演精度不高的问题,研究了利用北斗卫星导航系统B1、B2频段的信噪比数据进行土壤湿度的反演。提出了一种利用随机森林(Random Forest,RF)方法对B1、B2频段的观测量进行融合反演的方法,并利用粒子群优化(Particle Swarm Optimization,PSO)算法对RF的参数进行自动寻优。构建了PSO-RF土壤湿度反演模型,给出了信号处理的一般流程,并搭建陆基接收平台进行了验证实验。验证结果表明,PSO-RF模型反演结果较传统回归方法在相关系数R和均方根误差(Root Mean Square Error,RMSE)方面均有明显改善,证明了该方法可以实现对固定区域土壤湿度的长期连续观测。

Soil moisture monitoring based on GNSS-IR has overcome the shortcomings of traditional measurement methods,and has become a hot spot in remote sensing research in recent years.In view of the problem that the previous research mostly used the observation data of Global Positioning System(GPS)and the inversion precision is not high,the inversion of soil moisture by using the signal-to-noise ratio(SNR)data of Beidou B1,B2 frequency is studied.A fusion inversion method of B1 and B2 band observations using random forest(RF)method is proposed.The particle swarm optimization(PSO)algorithm is applied to automatically optimize the parameters of RF.The soil moisture inversion model based on PSO-RF and the relevant signal processing flow are presented.Moreover,a ground-based receiving platform is built for verification test.The experimental results show that the inversion results of PSO-RF model are significantly better than those of traditional regression methods in terms of correlation coefficient R and root mean square error RMSE and the PSO-RF model can realize a continuous long-term observation of soil moisture in the fixed area.