Optimizing Hydrologic Models Using Multi-Temporal LiDAR

The high degree of spatial variability of snow in complex terrain remains the greatest hurdle for obtaining more accurate forecasts of snow water equivalent in the Western United States. Expanding on previous work using repeated manual snow survey transects, we employ four independent airborne LiDAR scans of a 1-km² study site at Walton Creek in northern Colorado at different moments during the accumulation period (snow-free, early season, mid-season, and max-SWE) to show that distribution patterns of snow depth maintain consistency from year to year. Though the LiDAR acquisitions occurred throughout different winter seasons, drifting and scouring locations remained relatively unchanged for the three snow-on surveys. By normalizing and standardizing each of the three datasets of snow depth we determined a correlation cutoff value for extracting drift and scour locations from the remainder of the study area. Next, we used the known drift and scour locations to optimize and tune the terrain parameters of the wind redistribution model component of Isnobal, a physically-based, energy- and mass-balance snow model developed by the USDA-ARS. Finally, the optimal cutoff values for the wind model terrain parameters are applied to a nearby site with similar physiographic features to Walton Creek where changes in snow depth were estimated from the early- and mid-season LiDAR surveys. Early results reveal that this technique can predict broad areas of drift and scour, but struggles with very small-scale features, likely due to physical factors that are not accounted for by the simple wind model.