低植被覆盖环境下BDS双频多星组合反演土壤水分研究

Inversion of soil moisture by BDS dual-frequency multi-star combination under low vegetation cover environment

土壤水分是构成陆地生态系统的重要组成部分,高效、准确的监测其变化具有重要作用.利用全球导航卫星系统(Global Navigation Satellite System,GNSS)反射信号反演土壤水分已成为一种新型的遥感监测方法.目前,已有研究多倾向于在裸土环境下通过全球定位系统(Global Positioning System,GPS)数据反演土壤水分,同时在引入小波分析分离卫星反射信号时,缺乏对分解层数的探讨.为此,基于北斗卫星导航系统(BeiDou Navigation Satellite System,BDS),提出一种基于小波分析和最小二乘支持向量机的土壤水分双频多星组合反演方法.探讨不同卫星高度角下小波分解层数对卫星信号的分离效果,并对比分析不同双频卫星组合反演土壤水分的可行性和有效性.以低植被覆盖环境为例,实验表明:(1)由于观测数据受多种因素影响,不同卫星最佳的高度角范围和小波分解层数均不同.(2)采用最小二乘支持向量机能够有效综合各卫星不同频率信号包含的土壤水分信息,反演结果更准确;当双频组合为四星时,反演结果与参考值之间的相关系数为0.977,较单星单频提高了20.74%.双频多星组合不仅使模型反演的效果较好且稳定性更高,而且在低植被覆盖环境下也能保持较好的反演精度.因此,采用BDS双频多星组合反演土壤湿度是可行的.

Soil moisture is an important component that constitutes terrestrial ecosystems,and efficient and accurate monitoring of its changes plays an important role.Soil moisture inversion using Global Navigation Satellite System(GNSS)reflection signals has become a new remote sensing monitoring method.At present,most of the existing studies tend to invert soil moisture by Global Positioning System(GPS)data in bare soil environment,and at the same time,when introducing wavelet analysis to separate satellite reflection signals,the number of decomposition layers is not explored.Based on BeiDou Navigation Satellite System(BDS),this paper proposes a dual-frequency multi-satellite inversion method based on wavelet analysis and least squares support vector machine to investigate the effect of wavelet decomposition layers on the separation of satellite signals at different satellite altitude angles,and to compare and analyze the feasibility and performance of different dual-frequency satellite The feasibility and effectiveness of soil moisture inversion with different combinations of dual-frequency satellites were compared and analyzed.Taking the low vegetation cover environment as an example,the experiments show that:(1)the optimal altitude angle range and the number of wavelet decomposition layers of different satellites are different because the observation data are affected by various factors;(2)the least squares support vector machine can effectively integrate the soil moisture information contained in different frequency signals of each satellite,and the inversion results are more accurate;when the dual-frequency combination is four satellites,the correlation coefficient between the inversion results and the reference value is 0.977,which is 20.74%higher than that of single-satellite single-frequency.The dual-frequency multi-star combination not only makes the model inversion more effective and stable,but also maintains a better inversion accuracy under low vegetation cover environment.Therefore,it is feasible to use the BDS dual-frequency multi-star combination to invert soil moisture.