SnoLimits is a snow depth and SWE dataset developed by the Terrestrial Hydrology
group at CU Boulder. The initial random forest methodology was tested using 50 m
lidar snow depth data in Colorado basins and described in
Herbert et al. (2025).
For the SnoLimits dataset, we applied the methodology at 500 m resolution,
producing snow depth and SWE in mountainous areas of Colorado and California.
Data for 2001–2025 is available for download via
HydroShare.
We also upload near real time data to this website which is not available in the
HydroShare repository. This data is available with a two day lag period for
MODIS satellite data processing.
For more information, check out the cited papers, the project GitHub, or reach
out to us with any questions.
Mountainous regions in the western United States are key reservoirs of terrestrial water stored as snow, but limited observational coverage creates gaps in snow water equivalent (SWE) estimates. Snow stations provide real-time measurements at sparse points in space, while airborne lidar surveys capture spatially continuous snapshots at infrequent points in time. Together, these complementary datasets enable machine learning to learn the spatial patterns of snow from lidar while anchoring estimates to real-time ground conditions via snow stations. We present SnoLimits, a daily 500 m SWE and snow depth dataset spanning the MODIS era (2001–2025), created using a random forest model trained on physiographic and dynamic predictors. Spatial validation against withheld lidar surveys shows SnoLimits outperforms existing products (UASWE, ParBal, UCLA SWE), with lower RMSE and higher correlation (RMSE = 0.16 m, R² = 0.63 in Colorado; RMSE = 0.23 m, R² = 0.81 in California). Temporal validation at snow stations indicates performance comparable to UCLA SWE. SnoLimits is intended for hydrological modeling and water resource applications in Colorado and California, and could be made operational with a latency of a few days.