Observation, Simulation, and Evaluation of Snow Dynamics in the Transitional Snow Zone
Wayand, Nicholas E.
MetadataShow full item record
The frequent mid-winter accumulation and ablation cycles of snowpack within the rain-snow transitional zone play an important role for the maritime basins along the western U.S. mountain ranges. Representation of transitional snowpack within hydrological models has remained a challenge, largely because surface and meteorological conditions frequently remain near the freezing point, which allows large errors in modeled accumulation or ablation to result from small forcing or structural errors. This research aims to improve model representation of accumulation and ablation processes by utilizing new observations within the transitional snow zone combined with novel methods of model evaluation. The importance of mid-winter snowmelt during historical flooding events was assessed over three maritime basins in the western US. A physically-based snow model was coupled with an idealized basin representation to quantity how the characteristics of each basin combined with storm strength to control the distribution of snowmelt over a basin. Snowmelt contributions to total basin runoff ranged from 7-29% during historic flooding events between 1980 and 2008. However, poor meteorological forcing data were found to be a major limitation in model evaluation. In response to this limitation, a historical snow study site at Snoqualmie Pass within the Washington Cascades was updated in October 2012 with meteorological, soil, and snow observations to provide an ideal site for model evaluation within the transitional snow zone where existing observations are extremely sparse. The data set includes complete meteorological forcing required for snow models: air temperature, total precipitation, wind speed, specific humidity, air pressure, short- and longwave irradiance. Historical (1980-2015) observations include snow board new snow accumulation, multiple measurements of total snow depth, and manual snow pits, while more recent years (2012-2015) include sub-daily surface temperature, snowpack drainage, soil moisture and temperature profiles, and eddy co-variance derived turbulent heat flux; in short an ideal site to test different hypothesis about snow processes. This unique observational data set was used to illustrate how a novel process-based approach can diagnose model errors in snow accumulation processes (precipitation partitioning, new snow density, and compaction). The main source of model error on each day was identified by comparing observed snow board measurements to a “modeled snow board.” Results found that even after in-situ calibration, new snow density errors were the most common, occurring 53% of available days, followed by precipitation partition errors (43%) and compaction errors (18%). Daily errors canceled out on annual time scales during all years except the anomalously warm winter of 2014-2015. The partitioning of precipitation into rain or snow during water year 2015 was further examined by evaluating surface-based and mesoscale-model-based predictions. Observations of precipitation phase from a disdrometer at Snoqualmie Pass and nearby snow depth sensors were used to evaluate both methods. With calibration, the skill of surface-based methods was greatly improved by using air temperature from a nearby higher-elevation station, which was less impacted by surface inversions at the pass. Without any form of a prior calibration, we found a hybrid method that combines surface-based predictions with output from the Weather Research and Forecasting mesoscale model, to have comparable skill to calibrated surface-based methods. These results suggest that phase prediction in mountain passes can be improved by incorporating observations or models of the atmosphere aloft.
- Civil engineering