Using a Bayesian Regression Approach on Dual-Model Windstorm Simulations to Improve Wind Speed Prediction

Abstract

Weather prediction accuracy is very important given the devastating effects of extreme-weather events in recent years. Numerical weather prediction systems are used to build strategies to prevent catastrophic losses of human lives and the environment and have evolved with the use of multimodel or single-model ensembles and data-assimilation techniques in an attempt to improve the forecast skill. These techniques require increased computational power (thousands of CPUs) because of the number of model simulations and ingestion of observational data from a wide variety of sources. In this study, the combination of predictions from two state-of-the-science atmospheric models [WRF and RAMS/Integrated Community Limited Area Modeling System (ICLAMS)] using Bayesian and simple linear regression techniques is examined, and wind speed prediction for the northeastern United States is improved using regression techniques. Retrospective simulations of 17 storms that affected the northeastern United States during the period 2004–13 are performed and utilized. Optimal variances are estimated for the 13 training storms by minimizing the root-mean-square error and are applied to four out-of-sample storms [Hurricane Irene (2011), Hurricane Sandy (2012), a November 2012 winter storm, and a February 2013 blizzard]. The results show a 20%–30% improvement in the systematic and random error of 10-m wind speed over all stations and storms, using various storm combinations for the training dataset. This study indicates that 10–13 storms in the training dataset are sufficient to reduce the errors in the prediction, and a selection that is based on occurrence (chronological sequence) is also considered to be efficient.

Assessment of Storm Wind Speed Prediction Using Gridded Bayesian Regression Applied to Historical Events With NCAR's Real‐Time Ensemble Forecast System

“…BLR is implemented by combining 10 NCAR ensemble members, similar to the method applied in Yang et al () that combined output from two NWP models for a different set of historical storms than the ones considered herein. A Bayesian approach is used to deal with uncertainty of the unknown regression coefficients and variances for the ensemble members, and these unknown parameters are basically inferred to probabilistic forms.…”

“…We follow the assumption introduced by Yang et al () that the best postprocessed model will be the ensemble mean, indicated by a mean vector of μ β = [0 0.1…0.1]. If the prior variances ( ) are made smaller, the results will shrink toward this assumption.…”

“…This study proposes a new method of producing gridded, statistically postprocessed forecasts based on dynamic ensemble forecasts and provides another alternative to the traditional ensemble mean. Specifically, this work extends the Bayesian linear regression technique (BLR), a new “implicit” method using regression coefficients that combines all ensemble members and observations to produce a deterministic forecast (Yang et al, ). This methodology takes advantage of optimized regression coefficients instead of constraint weights used to create an average of ensemble members and enables improvements upon the “implicit” way for cases when all ensemble members are biased consistently either positive or negative.…”

“…The newly developed GBLR product is produced from real‐time, high‐resolution ensemble forecasts provided by the National Center for Atmospheric Research (NCAR; Schwartz et al, , ) and applied to instantaneous 10‐m above ground level wind speed forecasts. We would like to note that even though there are conventional and more popular Bayesian schemes available, we opted to use our own for the application with the NCAR ensemble system since it works very well for the prediction of storms in the NE United States (Yang et al, ). However, a comparison of BLR with the Bayesian horseshoe is provided to illustrate commonalities between the different techniques.…”

Study on Wind Simulations Using Deep Learning Techniques during Typhoons: A Case Study of Northern Taiwan

“…The ML model and statistical methods are based on myriad historical data not considering the meteorological conditions. Data-driven models are generally accepted as a feasible approach for establishing the capability for wind-speed predictions [24][25][26][27][28][29][30]. For instance, Li et al [31] proposed a data-driven model for determining the wind-power spectrum of a The numerical weather prediction (NWP) model is a method of weather simulation that employs a set of equations that describe the flow of fluids.…”