Comparative Analysis of AI-NWP and Dynamically-Driven Models for Snowfall Forecasting in the Rocky MountainsAngerman, Julia ForecastingAI-NWPOperational MeteorologyMeteorology and Atmospheric ScienceB SThe growing number of Artificial Intelligence (AI) forecasting models within the meteorological community prompts discussion of whether their performance is comparable to the traditional dynamical-driven, or Numerical Weather Prediction (NWP), models. NWP utilizes complex mathematical equations for model outputs, which have been used for decades in operational forecasting. While recent research on AI-NWP models, which are data-driven products trained on historical observations, have found instances in which they were superior to existing models, there has been little investigation that directly focuses on snowfall. This study examines 11 prominent snowfall events from 2022-2024 across the domain of Colorado for the variables of 2-meter temperature, u- and v-wind, 500 mb heights, mean sea level pressure, and accumulated precipitation at lead times from 0-240 hours. Statistical measures of root mean square error and threat score were used to analyze the performance of the GraphCast and GFS models. The findings suggest that the GraphCast generally met or exceeded GFS’ performance. These results provide a useful conclusion for operational forecasters that AI-NWP models can be used alongside existing dynamical models to create the most accurate forecast possible. Their performance is only expected to strengthen in the coming years, suggesting that further study of the products could yield daily usage in an operational setting. Steven J Greybush, Thesis SupervisorSteven J Greybush, Thesis Honors AdvisorKyle Alexander Imhoff, Faculty Readeropen_access2026-04-08T00:33:31Zhttps://honors.libraries.psu.edu/catalog/10096jka5627