Streamflow in the Columbia River Basin: Quantifying Changes over the Period 1951-2008 and Determining the Drivers of Those Changes

Trend and detection and attribution analyses were performed using naturalized streamflow observations and routed land surface model simulations for the Columbia River Basin (CRB) covering the water years 1951 – 2008. The CRB was separated into 10 subbasins with the flow from the Lower Columbia subbasin at The Dalles representing the flow for the entire basin. All the subbasins had significant declines in the amount of annual total streamflow except Middle and Upper Snake and Upper Columbia. These declines in annual total flow were led by significant declines in the monthly flow for June – October. Declines in June – October also directly led to significant declines in the peak flow and July-August-September summer means. Trends for center of timing were not as consistent across all the subbasins, but a significant shift towards earlier center of timing was found at the Lower and Upper Columbia and Kootenay subbasins. The Routing Application for Parallel computatIon of Discharge (RAPID) routed semi-factorial simulations of the land component of DOE Energy Exascale Earth System Model (ELM) were driven by three different sets of meteorological drivers with the temperature and precipitation corrected by Livneh (Livneh et al., 2013). The mean of the three sets of simulations provided the historical changes in streamflow due to the effects of climate change (CLMT), CO₂ concentration (CO2), nitrogen deposition (NDEP), and land use and land cover change (LULCC). Excluding the Snake River subbasins, LULCC had the same pattern of declines in monthly flow, but the period was shifted to May – September. The June – October pattern of significant trends was also found in NDEP; however, the trends showed significant increases in flow. While there were significant trends in CO2, NDEP, and LULCC, the detection and attribution analysis showed that the change in annual total, center of timing of, and summer mean streamflow could only be attributed to CLMT. This is due to the signals in CO2, NDEP, and LULCC being weak in comparison to the signal in CLMT and the natural internal variability found in streamflow. The results indicate that the historical changes in climate led to a decrease in summer flow for the CRB which consequently led to a decrease in annual total flow and a shift towards earlier center of timing.