Abstract This paper develops a simple method of predicting sediment removal during snowmelt and flashfloods in Alkali Creek watershed, western Colorado. The method links successful runoff simulations to field-monitored sediment relationships to predict site yield. The yield data are differenced to derive an index of “scour and fill”, from which it is possible to infer the patterns of erosion and deposition that the stimulated events might produce across the watershed. Site sediment yield patterns produced suggest that even for a basin with this south-facing orientation, the summer storms are geomorphologically ineffective when compared to the amount of sediment transport potentially associated with the melt flows from a stimulated snowpack. Predicted patterns of “scour and fill” are spatially and temporally variable, and are dependent on sediment supply laterally to the network. Scour is extensive at tributary junctions, explaining the overall network concavity. These patterns conform generally with monitored channel change in the network over a 13 year period, and can be contrasted with patterns of sediment yield documented for other climatic types. It is concluded that in this alpine semi-arid environment, the sediment-rich debris removal associated with the decay of a considerable snowpack in the spring months is the most important geomorphic process, and that tributary junctions appear likely to be the most active sites of change. Implications for threshold behaviour in these environments are also explored.