Evaluating Numerical Weather Prediction Forecasting Accuracy in Columbus, Ohio

The use of numerical weather prediction models to generate daily weather forecasts is ubiquitous. Numerical weather prediction utilizes complex sets of equations and computer algorithms in order to simulate the state of the atmosphere at some point in the future. Accurate weather forecasts are essential to everyday life; operational meteorologists rely on them for daily forecasts, the power industry uses them to plan electricity generation, and many other uses.
However, the complexity of the atmosphere can make it difficult to aggregate all inputs and produce a consistently accurate forecast. In order to ensure forecasts are representative of the atmosphere, it is important to look back and compare forecasts against measured data. Because calculations are only executed at specific points, it is not computationally feasible to have infinitely many points, creating challenges with model resolution. To acquire forecasts between
points, the models utilize interpolation which can create room for error. This investigation will perform a case study within Columbus, Ohio by deploying a weather station equipped with a thermometer, tipping bucket rain gauge, and cup anemometer. This study will gather temperature, precipitation, and wind data at 00Z, 06Z, 12Z, and 18Z each day and compare the data at the specified times to 3 major numerical weather prediction models: Global Forecasting
System (GFS), North American Mesoscale (NAM), and High Resolution Rapid Refresh (HRRR). Through this process, the study will determine which forecast model and app are the most accurate to the true, measured observations made by our weather station. This will better enhance the forecasting skills of meteorologists and provide clarity on which model is most accurate in the Columbus region.