Optimization of binary regression tree models for estimating the spatial distribution of snow water equivalent in an alpine basin

Regression tree models have been shown to provide the most accurate estimates of distributed SWE when intensive field observations are available. This work presents an approach that improves regression tree model performance by optimizing the use of independent variables and by comparing different residual interpolation techniques. The analysis was performed in the 19.1 km2 Tokopah basin, located in the southern Sierra Nevada of California. Snow depth, the dependent variable of the statistical models, was derived from three snow surveys (April, May and June, 1997), with an average of 990 depth measurements per survey. Estimates of distributed SWE were derived from the product of the snow depth surfaces, the average snow density (54 measurements on average), and the fractional snow covered area (obtained from the Landsat Thematic Mapper and the Airborne Visible/Infrared Imaging Spectrometer). Inclusion of the independent variable northness improved regression tree model fit. Co-kriging with solar radiation proved to be the best method for distributing residuals for April and June, with inverse distance weighting providing the best result for May.