CONTROL ID: 1820815

TITLE: A Computationally Efficient Platform To Examine the Efficacy of Regional Downscaling Methods

AUTHORS (FIRST NAME, LAST NAME): Jonathan L Vigh1, Caspar M. Ammann1, Richard B Rood1, Joseph J Barsugli1, Galina Guentchev1

INSTITUTIONS (ALL): 1. National Center for Atmospheric Research, Superior, CO, United States.


A primary goal of the National Climate Predictions and Projections (NCPP) platform is to develop an extensive set of standardized and interoperable evaluation tools to examine the efficacy of various regional climate downscaling techniques. To this end, a highly efficient NCL-based evaluation and comparison engine has been developed to compute period statistics on the monthly, seasonal, and annual timescales from many gridded daily datasets. The initial implementation of the engine computes metrics such as the mean, median, interannual standard deviation, monthly lag-1autocorrelation, and the various percentiles. These are computed for the daily maximum and minimum temperature, the daily mean temperature, the diurnal temperature range, and daily accumulated precipitation. Additionally, the engine computes the mean, median, max, and min of a number of application-oriented indices and climate extreme indices on the monthly and annual timescales. Output includes not only plots of the above metrics and indices, but the underlying NetCDF dataset used to create each plot. Metadata and internal attributes provide full data provenance of the source datasets and a description of the computations. Through this standardized approach to statistical analysis of massive datasets, we aim to provide end-users with a comprehensive suite of evaluations and comparisons between downscaling methods. This will enable applications-oriented users greater access to these data by lowering the barriers to data usage and understanding.

KEYWORDS: 1637 GLOBAL CHANGE Regional climate change.