Towards improved snow water equivalent retrieval algorithms for satellite passive microwave data over the mountainous basins of western USA

Space-borne passive microwave snow water equivalent (SWE) retrieval algorithms are attractive for continuous SWE monitoring over large mountainous areas. The performance of three SWE retrieval algorithms, which were considered relevant for operational purposes, was examined for each month over the Colorado River Basin. In addition, statistical post-processing was tested as a means of improving the SWE estimates from each algorithm. The evaluation started with the so-called Chang equation, which was a pioneer algorithm and is still used in practice. Successive attempts were then made to improve the algorithm's performance through the calibration of the equation's coefficient and through the inclusion of brightness temperature data from various frequency channels. The Chang equation consistently underestimated SWE with average bias between 30mm in November and more than 300mm in April and root mean square error (RMSE) exceeding 500mm at many locations in April. The statistical post-processing effectively removed the bias and reduced the RMSE by half for all the months. When the Chang equation's coefficients were calibrated at each site, biases were reduced by approximately 85%, and RMSE was reduced by 40%-50%. Finally, the multiple channel equations produced unbiased SWE estimates with RMSEs 50%-60% of those from the Chang equation. However, the statistical post-processing did not reduce RMSE for both calibrated algorithms. The last algorithm produced the most reliable estimates for at-site analysis, but its skill deteriorated when analyses were performed over larger areal extents; therefore, it is only recommended for SWE monitoring over smaller areas. For larger areas, the calibrated Chang equation is desirable because it only requires interpolations of a calibrated coefficient, which was spatially coherent.