Climate scenarios from global climate models (GCM) provide the primary scientific basis for our understanding of future climates. However, output data from GCMs are at coarse scales (i.e. 100-300 km) resulting in significant biases in regions of complex climate such as the Northwestern US making direct application of such data intractable. To overcome these limitations, GCM output must be translated (or downscaled) to local scales for modeling and value-added decision-making.
The Multivariate Adaptive Constructed Analogs (MACA) (Abatzoglou, Brown, 2012; Hegewisch, Abatzoglou, in prep.) method is a statistical downscaling method. MACA uses an observationally based training dataset that provides a basis for which the GCM is downscaled to. MACA is a multi-process approach that includes correcting for biases in GCM output and spatially downscaling. The latter is accomplished uses a set of historical analogs from observational data that capture many of the characteristics of regional climate.
MACA was used to downscale the model output from 20 global climate models (GCMs) of the Coupled Model Inter-Comparison Project 5 (CMIP5) for the historical GCM forcings (1950-2005) and the future Representative Concentration Pathways (RCPs) RCP 4.5 and RCP8.5 scenarios (2006-2100) from the native resolution of the GCM to either 4-km or ~6-km.
There are currently 3 CMIP5 MACA data products: MACAv1-METDATA, MACAv2-METDATA and MACAv2-LIVNEH.
MACAv1-METDATA is available for the Western US (and only at monthly timescales, while MACAv2-LIVNEH/MACAv2-METDATA are available over the entire coterminous USA.
MACAv2-LIVNEH/MACAv2-METDATA both the newest version of the MACA method (version 2), while MACAv1-METDATA uses version 1.
MACAv2-METDATA/MACAv1-METDATA utilized the METDATA training data so are on a 4-km (1/24-deg) grid,while MACAv2-LIVNEH utilizes the Livneh training data so is on a ~6-km (1/16-deg) grid.
Figure: The downscaling process is a translation of the information at a coarse resolution to a much finer resolution.
The MACA dataset is unique in that it downscales a large set of variables making it ideal for different kinds of modeling of future climate (i.e. hydrology, ecology, vegetation, fire, wind). We currently have data for the following variables:
tasmax - Maximum daily temperature near surface (2 m)
tasmin - Minimum daily temperature near surface (2 m)
rhsmax - Maximum daily relative humidity near surface (2 m)
rhsmin - Minimum daily relative humdity near surface (2 m)
huss - Average daily specific humidity near surface (2 m)
pr - Average daily precipitation amount at surface
rsds- Average daily downward shortwave radiation at surface
was - Average daily wind speed near surface (10 m)
uas - Average daily eastward component of wind near surface (10 m)
vas - Average daily northward component of wind near surface (10 m)
The MACA website has the following to offer:
Information: GCMs, the CMIP5 RCP climate experiments, the MACA method, the different MACA products
Data Portal: a data portal that aids in the download of CSV files for time series of point locations, netCDF subsets of rectangular regions of MACA data and the full data files.
Visualizations: spatial maps of future projections, scatter plots of future projections for a region
Guidance: advice on GCM selection, do/don'ts of analyzing climate data, links to external guides
Abatzoglou J.T. and Brown T.J. A comparison of statistical downscaling methods suited for wildfire applications. International Journal of Climatology (2012) doi: 10.1002/joc.2312.
Hegewisch,K.C., Abatzoglou J.T. An improved Multivariate Adaptive Constructed Analogs(MACA) Statistical Downscaling Method. In preparation.
Taylor, K.E., R.J. Stouffer, G.A. Meehl: An Overview of CMIP5 and the experiment design. MS-D-11-00094.1, 2012.
Katherine is a postdoctoral fellow working under Dr John Abatzoglou, a climatologist in the Department of Geography at the University of Idaho. Katherine's background is in computation and statistics. Katherine has been working on the statistical downscaling of global climate ...