Soil moisture monitoring is widely used in agriculture, water resource management, and disaster prevention, which is of great significance for sustainability. The global navigation satellite system interferometric reflectometry (GNSS-IR) technology provides a supplementary method for soil moisture monitoring. However, due to the quality of the signal-to-noise ratio (SNR) measurements and the complex surface environment, inevitable outliers in multipath interference signal metrics (amplitude, frequency, and phase) were used as modeling variables to inverse GNSS-IR soil moisture. Besides, it is hard to use the univariate model to comprehensively analyze the relationship between the various factors, due to the poor fitting effect and weak generalization ability of the model. In this paper, the minimum covariance determinant (MCD) robust estimation and machine learning algorithms are adopted. The MCD robust estimation can eliminate outliers of the multipath signal metrics and machine learning algorithms, including the back propagation neural network (BPNN), Gaussian process regression (GPR), and random forest (RF), and can comprehensively establish nonlinear GNSS-IR soil moisture inversion models using multipath interference signal metrics. Moreover, the study of the modeling parameter selection for the three machine learning algorithms and the inversion results for single satellite and all satellites are also carried out to make the algorithms more generalizable. The results show that the correlation coefficients (R) and the root mean square error (RMSE) of the machine learning models for all satellite tracks are increased by 4.3~86.6% and reduced by 2.8~30%, respectively, compared with the MCD multiple regression model. The RF model with 80 decision trees and 1 node shows the clearest improvement. The total model using all satellite data has more generalization ability than the single satellite model but causes some loss of accuracy.

For high spatiotemporal resolution global navigation satellite system interferometric reflectometry (GNSS-IR) snow depth monitoring, addressing terrain effects and multi-GNSS data fusion together is essential due to their coupling. Analyzed multi-GNSS spatiotemporal availability to ensure coverage and revisit rate. Improved data usage and inversion accuracy through Complete Ensemble Empirical Mode Decomposition (CEEMD). Corrected anisotropic terrain errors using DEM to account for varied reflection footprints. A grid was established for data partitioning and fusion. Considering inter-system errors, the proposed signal peak ratio weighting (PRW) fusion of single-system inversions is based on signal quality. Then, by using inverse distance weighting, the multi-GNSS results were fused, achieving high-accuracy, hourly snow depth inversions with high spatial resolution. With terrain correction, the Correlation Coefficient (R) reached 0.984, Root Mean Square Error (RMSE) 0.136m, and Mean Error (ME) -0.060m, reduced by 9.05% and 24.84%. PRW further enhanced accuracy, increasing R to 0.985, reducing RMSE 14.6% to 0.128m, improving ME 40.57% to -0.047m. Grid fusion effectively integrated multi-GNSS data, showing daily R 0.865, RMSE 0.102m, ME -0.050m. Across season, R 0.984, RMSE 0.134m, ME -0.065m. Compared to equal weighting, R improved 4.72% and 3.05%, RMSE reduced 26.09% and 14.10%, ME decreased 28.57% and 16.44%. Hourly results achieved 94.44% coverage, averaging 5.19 usable tracks, demonstrating effectiveness. Overall, this paper presents an end-to-end solution for high spatiotemporal resolution snow depth inversion using GNSS-IR, and the methodology can be extended to other geophysical parameter retrievals.