Data Sources

Global Climate Models

A Global Climate Model (GCM) is a type of General Circulation Model that focuses on projections of climate change by simulating how Earth’s physical processes respond to increasing greenhouse gas concentrations. Research groups worldwide develop GCMs for use in periodic climate assessment reports published by the United Nations Intergovernmental Panel on Climate Change (IPCC). GCM outputs help form the basis for many interpretations of future climate. The IPCC Fifth Assessment Report (AR5) was published in 2014.

For statistical and dynamical downscaling, SNAP uses:

  • CMIP5 model outputs from the IPCC Fifth Assessment Report (AR5). SNAP evaluated the performance of 22 GCMs used in CMIP5. Outputs include the first ensemble model run under historical and RCP 4.5, 6.0, and 8.5 climate scenarios
  • (Summer 2022) We continue to evaluate the performance of various GCMs for CMIP6. For several regions across Alaska and the Arctic, we are calculating and evaluating the degree to which each model’s output aligned with 1958–2001 climate data for precipitation, sea level pressure, and surface air temperature.

Historically gridded data

Historically gridded data are first directly observed or collected at a weather station or from a weather observer. Data are then interpolated to a common grid to fill in spatial gaps between observed points.

The number of variables from historically gridded data is small due to the costs associated with direct observations across a large enough spatial extent to warrant interpolation as a feasible technique to fill in the gaps. Historically gridded data are often used to validate model output, but there are often errors in the collection methods or in the assimilation of observations which need to be accounted for.

SNAP draws on historical gridded data from CRU and PRISM.

  • Climatic Research Unit (CRU)—The University of East Anglia’s Climatic Research Unit (CRU) provides monthly climate data from thousands of terrestrial and marine temperature stations. SNAP uses CRU TS data and CRU CL high resolution climatology data in our delta downscaled products.
  • PRISM—Parameter elevation Regression on Independent Slopes Model (PRISM) data are some of the highest resolution spatial climate data currently available across large extents. SNAP uses PRISM temperature and precipitation data from the North and other regions as the baseline climatology in its delta downscaled products. For statistical downscaling, SNAP uses:
    • temperature and precipitation data from the 30-year (1961–1990) monthly climatology at 2-km spatial resolution covering Alaska and regions of Canada
    • 771-m spatial resolution data from 1971–2000 covering only Alaska
    • other PRISM datasets for specific projects

Historical reanalysis data

Historical reanalysis data are generated from a weather model that tries to replicate the weather patterns of the past across the entire spatial extent of interest. This approach relies upon our understanding of how regional and global weather patterns work.

Reanalysis data can output a long list of internally consistent variables at sub-daily temporal resolutions. Limitations include coarser spatial resolutions (~80-km) depending on the type of model used and abrupt shifts in climate values.

SNAP uses historical reanalysis data from the European Centre for Medium-Range Weather Forecasts (ECMWF). Reanalysis produces global numerical weather forecasts for users worldwide. SNAP uses ERA-40 for model selection methods, and ERA-Interim for dynamical downscaling.