A general snow accumulation and melt model was developed to (1) determine how accurately snow accumulation and ablation can be modeled over heterogeneous landscapes with routinely available climatologic, topographic, and vegetation data, and (2) improve estimates of annual forest snow hydrology for point and regional calculations of annual forest productivity. The snow model was designed to operate within the Regional Hydroecological Simulation System (RHESSys), a GIS based modeling system to manage spatial data for distributed computer simulations on watershed scales. One feature of the RHESSys Snow Model (RSM) is it can use satellite derived forest leaf area index (LAI) to represent catchment forest cover; difficult to obtain in adequate cover and resolution by any other means. The model was tested over 3 water years (October to September) with data recorded by 10 snow telemetry stations (SNOTEL) in 5 states ranging in meso-climate and elevations from a coastal Oregon site (1067 m) to a continental Colorado site (3261 m). Predictions for the 10 sites were made with identical parameter values and only site climate varied for all sites. The average difference between observed and predicted snow depletion dates for all sites and water years was 6.2 days and 8 of the 30 simulations were within ±2 days (R2 = 0.91). Radiation melt was the dominate snow ablation component at the Colorado site where sublimation was 10% (LAI = 0) to 20% (LAI = 6) of snow loss while air temperature was the dominate component at the Oregon site with sublimation reduced to 1% (LAI = 0) to 6% (LAI = 6) of snow loss. LAI had a greater effect determining snow depletion than site aspect. Aspect increased in importance if the snow depletion occurred during early spring when solar insolation differences between hillslopes is greater than in the late spring. An accurate prediction of daily snowpack water equivalent (SWE) was not a strong determinant for making an accurate prediction of snowpack depletion date. Predicted snowpack depletion dates were more sensitive to timing when the snowpack reached an isothermal condition. Daily estimates of SWE were most sensitive to correctly estimating snowfall from SNOTEL data. This means that for purposes of determining the snow depletion dates which are useful for forest ecosystem modeling, tracking SWE is less important then triggering snowmelt. Comparisons of simulations to published snow depletion dates show that RSM predicted the relative ranking and magnitude of depletion for different combinations of forest cover, elevation, and aspect.
- Regional simulation
- Remote sensing