Assessing HRRR Representation Accuracy of Autumn Near-Surface Conditions over Ohio

Numerical weather prediction models are one of the primary tools used in operational forecasting. These forecasts are essential in making agricultural, transportation, and energy production decisions. The High-Resolution Rapid Refresh model (HRRR) is an hourly updated forecast model designed for regional-focused forecasting. The accuracy of the HRRR’s near-surface conditions is vital to forecasters in regions with more significant variability in climate. The state of Ohio is one such region with up to four different Köppen climate classifications due to its proximity to Lake Eerie, the Appalachian mountains, and the Ohio River Valley. Assessing the HRRR accuracy of the near-surface conditions over Ohio would assist forecasters in determining potential bias and misrepresentations that can result in inaccurate forecast products. This study evaluates the HRRR model’s ability to represent variables in near-surface conditions, including mean sea level pressure, temperature, and wind speed during autumn. The comparative analysis will be conducted using statistical methods for the model output and the Ohio Automated Surface Observing Systems (ASOS) network. Statistical analysis will assess model bias; root means square error and regression analysis. The study will examine and compare the four climate regions and the representation of the variability of the diurnal cycle. Insights from this study will help determine ways to improve the HRRR model in future updates and highlight the importance of accurate forecasting in regions with climate diversity.