Localized Trend Analysis of Multi-Model Extremes Metrics for the Fourth National Climate Assessment

We report on localized analysis of scenario-based datasets for the Fourth National Climate Assessment (NCA4). These datasets include CMIP5-based Localized Constructed Analogs (LOCA) downscaled simulations at daily temporal resolution and 1/16^th-degree spatial resolution. Over 30 temperature and precipitation extremes metrics have been processed using LOCA data, including threshold, percentile, and degree-days calculations.

The localized analysis calculates trends in the temperature and precipitation extremes metrics for relatively small regions such as counties, metropolitan areas, climate zones, administrative areas, or economic zones. For NCA4, we are currently addressing metropolitan areas as defined by U.S. Census Bureau Metropolitan Statistical Areas. Such localized analysis provides essential information for adaptation planning at scales relevant to local planning agencies and businesses.

Each locale is defined by a closed polygon that is used to extract LOCA-based extremes metrics specific to the area. For each metric, single-model data at each LOCA grid location are first averaged over several 30-year historical and future periods. Then, for each metric, the spatial average across the region is calculated using model weights based on both model independence and reproducibility of current climate conditions. The range of single-model results is also captured on the same localized basis, and then combined with the weighted ensemble average for each region and each metric. An example for cooling degree days (CDD) in the Los Angeles area is shown below.

We also discuss lessons learned in the development and application of these techniques, as well as their relevance to risk analysis for adaptation planning.