Comparison of Spatial Prediction Methods for Estimating Snow Distribution in the Colorado Rocky Mountains

Seasonally snow covered areas account for a significant percentage of the land surface of the earth. Seasonal snow cover is important due to its contribution to annual runoff and due to its role in defining the global energy balance. Our understanding of snow distribution in the mountains is limited as a result of the extreme spatial variability snow exhibits. More accurate representations of snow distribution are greatly needed for improvements to hydrological forecasts, climate models, and for the future testing and validation of remote-sensing retrieval algorithms. In this study, the relative performances of four spatial prediction methods were evaluated to estimate snow water equivalent (SWE) for three 1 km² study sites in the Colorado Rocky Mountains. Each study site is representative of different topographic and vegetative characteristics. From 1-11 April 2001, 550 snow depth measurements and approximately 16 snow density profiles were obtained within each study site. The analytical methods used to estimate snow depth over the 1 km² areas were 1) inverse distance weighting, 2) ordinary kriging, 3) modified residual kriging and cokriging, and 4) a combined method using binary regression trees and cokriging. The independent variables used were elevation, slope, aspect, net solar radiation, and vegetation. Using cross validation procedures, each method was assessed for accuracy. Snow density was modeled over the 1 km² area using a linear regression technique. Snow depth estimates from the "best" or most accurate method were combined with modeled snow depth to produce SWE estimates for each of the study sites.