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
The statistics of temperature and precipitation may vary in the future, and the Department of Defense (DoD) infrastructure may be adversely impacted by a change in the statistical properties of these variables. It is distinctly possible that the risk is correlated across multiple DoD facilities, but non-stationarity effects must be considered to assess this risk. Therefore, this research’s objective is to build the foundation for a comprehensive determination of this risk by using historical observations to inform model projections of the statistical properties of temperature and precipitation, including non-stationarity, at regional scales in the Continental United States (CONUS). This will yield the spatial and temporal structure and covariability of these fields and can enable the evaluation of the correlated risk of extreme temperature and precipitation events on DoD facilities.
The null hypothesis of this research is that the statistics of temperature and precipitation and their covariability in the CONUS on regional scales have been and will continue to be stationary. To test this hypothesis, the research team will use National Centers for Environmental Information (NCEI) weather station data hourly maps of temperature and precipitation and 3-hourly temperature and precipitation fields from the global circulation models (GCMs) covering 1981-2010 in conjunction with a product-sum covariance (variogram) model of anisotropic spatio-temporal domains to determine the stationarity of observational and modeled datasets. Researchers will evaluate the performance of the GCMs relative to historical observations of temperature and precipitation non-stationarity, and then use these results to produce, as a function of emissions scenario, temperature and precipitation fields that specifically capture observed covariance and non-stationarity. Finally, researchers will use these results to determine the risk of significant change in Köppen-Geiger class for a given emissions scenario for the CONUS at regional scales.
This research will provide robust estimates, by leveraging the information contained in historical observations and an ensemble of global circulation model projections, of (1) the risk of changes in the distribution of temperature and precipitation that a single DoD facility may face for a set of plausible emissions scenarios and (2) the risk that multiple DoD facilities may face simultaneously. This project has the potential to be adapted to explicitly evaluate the correlated risk of extreme temperature and precipitation events across multiple DoD facilities and can easily be downscaled to include facility-level datasets. For the scientific community, this research will pioneer the use of advanced spatio-temporal variography and kriging to blend historical data and GCM projections to examine the spatio-temporal behavior of temperature and precipitation over the CONUS in future emission scenarios.