An Improved Extreme Forecast Index for Temperature and Precipitation

Forecasters are increasingly overwhelmed by the fire hose of data available to them. It’s imperative that they have tools that help them efficiently and accurately distill important information from this data. This study presents an improved extreme forecast index for temperature and precipitation for situational awareness. First, temperature and precipitation were evaluated from several reanalyses across the complex terrain of British Columbia, Canada. The Japanese Meteorological Agency’s 55-year Reanalysis (JRA-55) was found to best represent both typical and extreme values of these variables. The JRA-55 was then downscaled and bias corrected using the Parameter Elevation Regressions on Independent Slopes Model (PRISM) and weather station observations to create an improved, very high resolution 60-year surface analysis (VHRSA). Next, the VHRSA was used to bias correct and downscale North American Ensemble Forecast System (NAEFS) forecasts, creating more skillful probabilistic forecasts. Finally, from this a new Parametric Extreme Index (PEI) was created — a situational awareness tool to alert forecasters to upcoming extreme values of temperature and precipitation. The PEI, which employs a distribution appropriate for extreme values, is shown to me more skillful than existing indexes that assume a Gaussian distribution, providing a more accurate assessment of how extreme an event is. Further, the PEI takes into account the nonstationarity of the distribution (i.e., warming climatological distribution) that was found to be significant for minimum daily temperature.