Quantifying Spatiotemporal and Elevational Precipitation Gauge Network Uncertainty in the Canadian Rockies
Abstract. Uncertainty in estimating precipitation in mountain headwaters can be transmitted to estimates of river discharge far downstream. Quantifying and reducing this uncertainty is needed to better constrain the uncertainty of hydrological predictions in rivers with mountain headwaters. Spatial estimation of precipitation fields can be accomplished through interpolation of snowfall and rainfall observations, these are often sparse in mountains and so gauge density strongly affects precipitation uncertainty. Elevational lapse rates also influence uncertainty as they can vary widely between events and observations are rarely at multiple proximal elevations. Therefore, the spatial, temporal, and elevational domains need to be considered to quantify precipitation gauge network uncertainty. This study aims to quantify the spatiotemporal and elevational uncertainty in the spatial precipitation interpolated from gauged networks in the snowfall-dominated, triple continental divide, Canadian Rockies headwaters of the Mackenzie, Nelson, Columbia, Fraser and Mississippi rivers of British Columbia and Alberta, Canada and Montana, USA. A 30-year (1991–2020) daily precipitation database was created in the region and utilized to generate spatial precipitation and uncertainty fields utilizing kriging interpolation and lapse rates. Results indicate that gauge network coverage improved after the drought of 2001–2002, but it was still insufficient to decrease domain-scale uncertainty, because most gauges were deployed in valley bottoms. It was identified that deploying gauges above 2000 m will have the greatest cost-effective benefits for decreasing uncertainty in the region. High-elevation gauge deployments associated with university research and other programs after 2005 had a widespread impact on reducing uncertainty. The greatest uncertainty in the recent period remains in the Nelson headwaters, whilst the least is in the Mississippi headwaters. These findings show that both spatiotemporal and elevational components of precipitation uncertainty need to be quantified to estimate uncertainty for use in precipitation network design in mountain headwaters. Understanding and then reducing these uncertainties through additional precipitation gauges is crucial for more reliable prediction of river discharge.
Viewed (geographical distribution)