Multi-variable evaluation of hydrological model predictions for a headwater basin in the Canadian Rocky Mountains

Abstract. One of the purposes of the Cold Regions Hydrological Modelling platform (CRHM) is to diagnose inadequacies in the understanding of the hydrological cycle and its simulation. A physically based hydrological model including a full suite of snow and cold regions hydrology processes as well as warm season, hillslope and groundwater hydrology was developed in CRHM for application in the Marmot Creek Research Basin (~ 9.4km2), located in the Front Ranges of Canadian Rocky Mountains. Parameters were selected from digital elevation model, forest, soil and geological maps, and from the results of many cold regions hydrology studies in the region and elsewhere. Non-calibrated simulations were conducted for six hydrological years during 2005–2011 and were compared with detailed field observations of several hydrological cycle components. Results showed good model performance for snow accumulation and snowmelt compared to the field observations for four seasons during 2007–2011, with a small bias and normalized root mean square difference (NRMSD) ranging from 40 to 42% for the subalpine conifer forests and from 31 to 67% for the alpine tundra and tree-line larch forest environments. Overestimation or underestimation of the peak SWE ranged from 1.6 to 29%. Simulations matched well with the observed unfrozen moisture fluctuation in the top soil layer at a lodgepole pine site during 2006–2011, with a NRMSD ranging from 17% to 39%, but with consistent overestimation of 7 to 34%. Evaluations of seasonal streamflow during 2006–2011 revealed the model generally predicted well compared to observations at the basin scale, with a NRMSD of 77% and small model bias (6%), but at the sub-basin scale NRMSD were larger, ranging from 86 to 106%; though overestimation or underestimation for the cumulative seasonal discharge was within 24%. Timing of discharge was better predicted at the Marmot Creek basin outlet having a Nash-Sutcliffe efficiency (NSE) of 0.31 compared to the outlets of the sub-basins where NSE ranged from −0.03 to −0.76. The Pearson product-moment correlation coefficient of 0.12 and 0.17 for comparisons between the simulated groundwater storage and observed groundwater level fluctuation at two wells indicate weak but positive correlations. The model results are encouraging for uncalibrated prediction and indicate research priorities to improve simulations of snow accumulation at treeline, groundwater dynamics and small-scale runoff generation processes in this environment. The study shows that improved hydrological cycle model prediction can be derived from improved hydrological understanding and therefore is a model that can be applied for prediction in ungauged basins.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.