Determining the Temporal and Spatial Scales of Nonstationarity in Temperature and Precipitation across the Continental United States for a Given Emissions Scenario

Daniel Feldman | Lawrence Berkeley National Laboratory

RC18-1577

Objective

The means and extremes of surface air temperature and precipitation have changed significantly in the last 50 years, and may change more in the 21st Century. These changes may be without a historical analogue, and this nonstationarity presents a challenge for Department of Defense (DoD) infrastructure planning and operations, wherein historical distributions are used as the basis for managing infrastructure and operations risk from flooding, drought, heat waves, and other weather phenomena. The objective of this project was to use a combination of climate models, observations, and analysis to develop robust, defensible projections of the range of distributions of temperature and precipitation that DoD infrastructure and operations will face.

Back to Top

Technical Approach

There are several important tools for the development of these projections: Global Circulation Models and downscaling. This investigation advanced the understanding of a framework to manage these challenges and then determined the proper approach(es) to translating the coarse and biased climate model projections into the fine-scale, calibrated projections. This work used ~30 models from the Coupled Model Intercomparison Project – Phase 5 (CMIP5) downscaled with the Localized Constructed Analogues technique and developed regressed projections of 21st Century climate across the Conterminous United States.

Back to Top

Results

The project team found that a regression-based statistical framework can be employed to develop these projections but they also found that there are significant pitfalls that must be addressed before these projections can be confidently utilized by end-users. There are a number of potentially significant physical mechanisms, including changes in snow and coast weather pattern shifts, which can violate stationarity assumptions, and a regression-based approach using statistically downscaled models would not capture these effects. Therefore, the project team found that it will be necessary to advance research that addresses these pitfalls by (1) specifically testing stationarity assumptions inherent in statistical downscaling techniques, and (2) undertaking a set of intercomparisons between statistical and dynamical downscaling solutions to uncover and quantify the physical mechanisms that violate stationarity assumptions. The results of these expanded research efforts can then serve as basis projections from the upcoming CMIP exercises (e.g., CMIP6) to ensure that they are appropriate for the development of best available estimates for how temperature and precipitation projections may change over the 21st Century.

Back to Top

Benefits

The benefit of this research was identifying what additional, targeted information is needed to ensure that local risks to DoD infrastructure and operations are not underestimated due to known physical mechanisms that could be un- or under-represented in climate models.

Back to Top

Points of Contact

Principal Investigator

Daniel Feldman

Lawrence Berkeley National Laboratory

Phone: 510-495-2171

Program Manager

Resource Conservation and Resiliency

SERDP and ESTCP

Share