Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Etats Unis → 2012 → Assessing the Value of Improved Snow Information in Operational Hydrologic Models

Boise State University (2012)

Assessing the Value of Improved Snow Information in Operational Hydrologic Models

Burnop Alison Christine

Titre : Assessing the Value of Improved Snow Information in Operational Hydrologic Models

Auteur : Burnop Alison Christine

Université de soutenance : Boise State University

Grade : Masters of Science in Hydrological Sciences 2012

The National Weather Service’s (NWS) operational hydrologic model, the Sacramento Soil Moisture Accounting Model (SACSMA), coupled with their temperature index snowmelt model, SNOW-17, were implemented in the Dry Creek Experimental Watershed (DCEW), located in the semi-arid region of southwestern Idaho, just north of the city of Boise, Idaho. The model was downscaled from the standard 1 HRAP to ¼ HRAP spatial resolution then calibrated using a modified manual calibration procedure from the NWS. The main modification was to decouple the SNOW-17 and SACSMA models during the calibration stage in order to accurately simulate the snow distribution without the interference of SACSMA parameter variations. The value of a site derived empirical areal depletion curve (ADC) in SNOW-17 was tested by comparing the calibrated model run using an empirical ADC to a model run using a calibrated ADC from the regional NWS office. The empirical ADC model run more accurately predicted the snow depth at internal watershed locations throughout the model simulation time period, 2000-2011. During these times, the streamflow leaving the outlet of the watershed was also more accurately simulated. The empirical ADC model run statistically performed better than the NWS calibrated ADC model run with lower RMSE, higher NSE, and more often had lower percent bias. Several of the dominant SNOW-17 parameters including the snowmelt temperature (MBASE), rain/snow temperature (PXTMP), and snow correction factor (SCF) produced more accurate snow accumulation and melt patterns as well as improved model output when they were allowed to vary spatially across the watershed. This spatial variance allowed the model to produce the correct distribution of observed snow depth in the higher elevation pixels.Several rain on snow (ROS) events occurred during the model simulation and the model responded differently depending on the severity of the ROS event and timing of temperature increase above freezing. This phenomena is simulated accurately during small events, 2-5 mm of rainfall, and during times when the temperature changes to above freezing in the middle of the precipitation event. When there are large rainfall events onto a snowpack during short periods of time, such as in the January 2011 ROS event, the model framework does not have the capability to handle these large events and underpredicts discharge. SNOW-17, by design, is very sensitive to large changes in air temperature, typically melting much more snow than is observed because of the models’ linear relationship between snowmelt and temperature on any given day. Model simulations compared to observed snow depth records in DCEW indicate that snowmelt does not increase linearly with increases in temperature.


Version intégrale

Page publiée le 9 novembre 2012, mise à jour le 12 juillet 2018