Climate change, largely a result of anthropogenic greenhouse gas emissions, is expected to lead to significant warming of the planet over the coming decades (IPCC, 2014). This increase in temperature will be accompanied by changes in other aspects of the climate system, such as atmospheric circulation and precipitation. Resulting changes in hydrological fluxes (i.e. streamflow, evapotranspiration) and states (i.e. snow water equivalent, soil moisture) are likely to change the flow regime of many rivers around the world. The Columbia River, whose flow regime is heavily dependent on seasonal snow melt, will likely experience significant changes in the timing of the streamflow and possibly in total flow volume.

In 2013, The Bonneville Power Administration (BPA) solicited proposals as part of its Technology and Innovation Program to develop “[…] a new set of temperature, precipitation, snowpack, and streamflow forecast projections for the entire Columbia River System, based on the new Global Climate Model datasets being published in conjunction with IPCC-5”, where IPCC-5 refers to the global climate model experiments performed in support of the fifth assessment report of the IPCC (2014). In addition to updated global climate models, BPA was interested in an evaluation of the effect of methodological choices in the modeling process on the range of projected future hydrological conditions. Other requirements were:

  • output at a daily time step for the period 1950-2100
  • at least two Representative Concentration Pathways (RCPs)
  • at least two downscaling techniques
  • at least two hydrological models to generate unregulated flows
  • account for glacial melt
  • provide streamflow projections that are unbiased in the 1950-2010 (historic) period relative to best-available, estimated natural streamflows
  • provided in a format usable by all three River Management Joint Operating Committee (RMJOC) members to run hydroregulation studies.

The work would be an update to the previous Columbia River climate change study performed for the RMJOC by the Climate Impacts Group at the University of Washington (Hamlet, 2011). Coloquially, this new iteration can be referred to as the Columbia River Climate Change study or CRCC.

The work presented here is the outcome of the project that was awarded to the UW Hydro | Computational Hydrology group at the University of Washington and the Oregon Climate Change Research Institute at Oregon State University. As detailed in the Methods section, the resulting data set takes advantage of advancements in climate models, downscaling methods, and hydrological modeling. The dataset is intended to be used for long-range planning by a variety of stakeholders in the region.

The general methodology that was used in the production of this data set follows that used in the production of the previous dataset (Hamlet, 2011). Specifically, we evaluated, downscaled and bias-corrected output from CMIP5 global models, so that the output could be used as input for regional-scale hydrological models, which cover the Columbia River Basin at a spatial resolution of 1/16º. Multiple hydrological models were then used to simulate the hydrology of the Columbia River Basin. The resulting spatial runoff was routed through the channel network to selected flow locations to produce daily streamflow sequences which were bias-corrected to produce daily natural flow records for the period 1950-2100. These streamflow time series are intended for use in impact studies. This procedure was applied to global climate model output based on two RCPs (RCP 4.5 and RCP 8.5). In accordance with the Opportunity Announcement, we produced transient streamflow time series for the entire 1950-2100 period.

Improvements and changes relative to the existing streamflow dataset (Hamlet, 2011) consist of the following:

  • latest CMIP5 global climate model output, which also form the basis for the IPCC AR5 climate change assessments, as well as recent regional climate modeling
  • multiple downscaling methods
  • multiple hydrological models
  • newer versions of GCMs, hydrologic models and streamflow bias-correction method.

While we have done our best to ensure consistency and quality across all permutations and locations, errors and/or problems do pop up on occasion. We keep track of known issues and their resolution.