Modern (1992–2011) and projected (2012–99) peak snowpack and May–July runoff for the Fort Peck Lake and Lake Sakakawea watersheds in the Upper Missouri River Basin

Abstract

Mountain snowpack is an important contributor to runoff in the Upper Missouri River Basin; for example, high amounts of winter and spring precipitation in the mountains and plains in 2010–11 were associated with the peak runoff of record in 2011 in the Upper Missouri River Basin. To project trends in peak mountain snowpack and runoff in the upcoming decades, multiple linear regression models of peak mountain snowpack and total May–July runoff were developed for the Fort Peck Lake (above Fort Peck Dam) and lower Lake Sakakawea watersheds (between Fort Peck and Garrison Dams) in the Upper Missouri River Basin. Input to regression models included seasonal estimates of precipitation, air temperature, and total reference evapotranspiration stratified by elevation. Calibration was based on records from 107 weather stations from 1991 to 2011. Regressed annual peak mountain snowpack was used as input to the transfer function of May–July runoff. Peak snowpack and May–July runoff were projected for 2012–99 on the basis of air temperature and precipitation from the Community Climate System Model (CCSM) output. Two estimates of projected peak snowpack and May–July runoff for 2012–99 were computed: one estimate was based on output from the CCSM, version 3.0 (CCSM3), and the second estimate was based on output from the CCSM, version 4.0 (CCSM4). The significance of projected trends was based on the Kendall’s tau nonparametric test. Annual peak snowpack was projected to have a downward trend for the Fort Peck Lake watershed and an upward trend for the lower Lake Sakakawea watershed. Projections of May–July runoff had a significant downward trend for the Fort Peck Lake, lower Lake Sakakawea, and Lake Sakakawea (combination of Fort Peck Lake and lower Lake Sakakawea) watersheds. Downward trends in projected May–July runoff indicated that power production at Fort Peck Dam might be affected particularly in the later part of the simulation (2061–99); however, confidence in projected May–July runoff for the later part of the simulation was less certain because bias-corrected air temperatures from CCSM3 and CCSM4 commonly fell outside of the observed range used for calibration. Projected May–July runoff combined for the Fort Peck Lake and lower Lake Sakakawea watersheds were on the order of magnitude of the 2011 flood for 1 simulation year for each of the CCSM-based simulations. High peak snowpack and precipitation in April, May, and June in the plains was associated with large May–July runoff events; therefore, high precipitation at lower elevations in the Fort Peck Lake and lower Lake Sakakawea watersheds was a factor in the simulation of extreme runoff events at the magnitude of the 2011 flood.