Assimilation of snow information into a cold regions hydrological model
Spring and summer snowmelt runoff from the Canadian Rocky Mountains recharge many rivers and hence provide critical water supplies for a large portion of the population in western Canada. Because of the complex topography and vegetation conditions, the sparse network of observations of climate and snow properties, and the low quality of atmospheric model products, data assimilation (DA) is a potentially useful tool to improve the forecasting and prediction of snow properties and streamflow. To achieve better snowpack and streamflow estimations using DA, this research aims to: 1) evaluate the usefulness of SNODAS SWE data in Canada, and determine the influence of processes missing from the SNODAS model on the accuracy of SNODAS SWE, 2) explore the possibility of using remotely sensed data for detecting snow interception in forest canopies, 3) assimilate in situ measured and remotely sensed snow interception data into CRHM and assess their influence on the simulation of snow interception losses, 4) determine the optimal method to assimilate in situ snow measurements into the CRHM for prediction of basin snowpacks and streamflow. The results illustrate: 1) missing snow processes (blowing snow transport and canopy snow interception and sublimation) in the SNODAS snow model contribute substantially to its overestimation of SWE, 2) canopy intercepted snow can be detected by optical remote sensing data (NDSI and NDVI), 3) automated snow depth data measured from an adjacent forest and clearing can be used in a mass budget to accurately quantify snow interception loss, and assimilation of in situ measured and remotely sensed snow interception information can all improve simulations of snow interception timing and magnitude, 4) assimilating in situ SWE and snow depth into CRHM generally improves the simulation of snowpack properties and streamflow, but the results varied among different assimilation schemes. A better SWE simulation through DA does not always lead to better prediction of streamflow. The advanced snow interception measurement and DA techniques presented here deepens the understanding of cold regions hydrological DA and improve the capacity to forecast and predict the hydrology of headwater river basins in the Canadian Rockies and other similar regions.
Data assimilation, Hydrology, Snow, SWE, Snow depth, Snow interception, SNODAS, Remote sensing
Doctor of Philosophy (Ph.D.)
Geography and Planning