Reforecasts: An Important Dataset for Improving Weather Predictions

Abstract

A "reforecast" (retrospective forecast) data set has been developed. This data set is comprised of a 15-member ensemble run out to two weeks lead. Forecasts have been run every day from 0000 UTC initial conditions from 1979 to present. The model is a 1998 version of the National Centers for Environmental Prediction's Global Forecast System (NCEP GFS) at T62 resolution. The 15 initial conditions consist of a reanalysis and seven pairs of bred modes.This data set facilitates a number of applications that were heretofore impossible.Model errors can be diagnosed from the past forecasts and corrected, thereby 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. 3 CAPSULE SUMMARY:Reforecasts (retrospective forecasts from the same model used operationally) can dramatically improve forecast skill through the statistical correction of the current forecast using the older forecasts.

The boreal spring variability of the Intra-Americas low-level jet and its relation with precipitation and tornadoes in the eastern United States

“…The main teleconnection associated with the IA-LLJ variability is the PNA. Prediction of the IA-LLJ variability may, therefore, be promising given that Hamill et al (2006) found the PNA to be one of the three most predictable patterns with a 10-day lead.…”

Rain-on-Snow Events in the Western United States

“…There is potential to improve flood forecasts by implementing snow data assimilation methods (McGuire et al 2006; Andreadis and Lettenmaier 2006;Slater and Clark 2006;Clark et al 2006), as well as innovative methods to use downscaled output from numerical weather prediction models (Clark and Hay 2004;Hamill et al 2006). However, improving forecasts of rain-on-snow floods is limited by the lack of general knowledge about the character of rain-on-snow events in the western United States.…”

Performance and reliability of multimodel hydrological ensemble simulations based on seventeen lumped models and a thousand catchments

“…Unreliable simulations can be misleading and should be used with caution, if at all. Many methods for post processing the probabilistic forecasts from ensembles have been proposed, such as the ensemble dressing (i.e., kernel density) approaches (Roulston and Smith, 2003;Wang and Bishop, 2005;, Bayesian model averaging , nonhomogeneous Gaussian regression , logistic regression techniques (Hamill et al, 2004(Hamill et al, , 2006, analog techniques (Hamill et al, 2006), forecasting assimilation (Stephenson et al, 2005), statistical postprocess calibration approach (Wood and Schaake, 2008), variance inflation method (Johnson and Bowler, 2009), the simple binning technique (Stensrud and Yussouf, 2007) and several others. However, these procedures were not considered in the present work.…”