Synoptic controls of the Colorado River basin snowfall are determined from 700 mb atmospheric circulation. The 700 mb time series is run through an S-mode Principal Component Analysis (PCA) which creates a synoptic index over the western US region This synoptic index is used as input to a feed-forward backpropagation neural network to develop transfer functions that simulate daily snowfall at approximately the 70% accuracy level. Thus, the neural network provides a scale translation mechanism for determining local surface parameters from synoptic scale circulation. The nature of the synoptic controls on the Colorado River basin snowfall are determined by iteratively removing information from the neural nets and determining the influence of specific atmospheric variance regions on snowfall. The primary synoptic control on the Colorado River basin snowfall comes from atmospheric variance in the Pacific Northwest reflecting synoptic systems that originate over the Gulf of Alaska in concert with cyclogenesis in the lee of the Rocky Mountains. Different Colorado River basin snow sub-regions, however, exhibit varying amounts of influence from the atmospheric variance regions defined by the synoptic index.