This dataset was created with the ClimateNA-ERA v6.40 software package for historical and future 30-year normal periods. At present, only the 1961-1990 Normal period is available for testing and review:

If you want to quantify uncertainty in future projections, you need to work with a selection of multiple, individual models. This is easily done with the provided software for locations of interest (see video tutorial), and in addition we provide gridded data for nine high quality and representative AOGCMs.

To help with the selection of a representative set of models for different regions of North America (or for the entire continent), we used the Katsavounidis-Kuo-Zhang (KKZ) algorithm, which selects an optimally representative set of future projections for different IPCC climate reference regions considering multiple climate variables (click on left panel dendrogram to visualize similarity among models):

A common approach is to select a median, a pessimistic and an optimistic projection, i.e., a subset size of 3. For example, to assess uncertainty of future projections in the Northwest North America region (NWN) with a 3-model ensemble, you would choose the models: CNRM (median), EC (optimistic), and ACC (pessimistic) from the table below (compare with the dendrogram above). More models (rows 4 to 8) will provide increasingly better represenation of uncertainty in future predictions in multivariate space.

To include the sensitive "outlier" model UKES in the model selection process, use the lower portion of the table. If UKES is included, this scenario will almost always be picked second, as the most "pessimistic" model of global warming. UKES is not a very likely outcome, but if you work with a larger ensemble, you may include it as one possible outcome. For details on the GCM selection process, see Mahoney et al. (2022).