While mountains provide essential water for crops and cities, published estimates of total water stored in major mountain ranges vary by a factor of five. Precipitation and temperature both change rapidly over complex terrain, and sampling is difficult in remote and rugged regions. Despite this complexity and uncertainty, seasonal skill in runoff forecasting from snow-covered mountains is much greater than for rain-dominated watersheds. Since the 1930’s, humans with snow depth probes have predicted spring and summer streamflow in the western United States. Despite dramatic changes in snow depth across both space and time, these index measurements have incredible skill because spatial patterns of snow accumulation are controlled by topography and repeat from one year to the next. Different spatial patterns control snowmelt and subsurface storage, but these also repeat each year. By studying these patterns, we can increase our predictive skills, but only if we are constantly vigilant to separate the good data from the bad, as mountains are renowned for breaking sensors, blocking radars, and baffling satellite-based sensors.