Reforecasts: An Important Dataset for Improving Weather Predictions

Abstract

A ?reforecast? (retrospective forecast) dataset has been developed. This dataset is comprised of a 15-member ensemble run out to a 2-week lead. Forecasts have been run every day from 0000 UTC initial conditions from 1979 to the present. The model is a 1998 version of the National Centers for Environmental Prediction's (NCEP's) Global Forecast System (GFS) at T62 resolution. The 15 initial conditions consist of a reanalysis and seven pairs of bred modes. This dataset facilitates a number of applications that were heretofore impossible. Model errors can be diagnosed from the past forecasts and corrected, thereby dramatically increasing the forecast skill. For example, calibrated precipitation forecasts over the United States based on the 1998 reforecast model are more skillful than precipitation forecasts from the 2002 higher-resolution version of the NCEP GFS. Other applications are also demonstrated, such as the diagnosis of the bias for model development and an identification of the most predictable patterns of week-2 forecasts. It is argued that the benefits of reforecasts are so large that they should become an integral part of the numerical weather prediction process. Methods for integrating reforecast approaches without seriously compromising the pace of model development are discussed. Users wishing to explore their own applications of reforecasts can download them through a Web interface.