Abstract

The demand for accurate, high-resolution severe weather forecasting continues to drive innovation in Convection-Allowing Models (CAMs). Over the past three years (2024–2026), the NASA Goddard Earth Observing System (GEOS) has undergone significant developments to operate at convective scales, culminating in successful evaluations at the NOAA/NWS Storm Prediction Center’s Hazardous Weather Testbed (HWT) Spring Forecast Experiments (SFE). This seminar highlights the evolution, evaluation, and operational successes of GEOS-CAM during these premier severe weather forecasting experiments. The presentation is divided into three parts: Part 1: Model Innovation and Grid Design (Bill Putman)The seminar will begin with an overview of the foundational model developments that brought GEOS-CAM to the convective scale. Bill Putman will detail the design and implementation of the 2-kilometer CONUS stretched grid, which allows for ultra-high-resolution forecasting over the contiguous United States while maintaining global context. This section will also highlight the convective-scale capabilities of GEOS featuring the FV3 dynamical core and showcase initial meteorological results that demonstrate the model's ability to accurately resolve severe convective structures. Part 2: Severe Storm Prediction: From Early Roots to Using CAMs and Machine Learning Tools at the HWT Spring Forecast Experiment (Gary Partyka) Nationwide severe storm prediction can be traced back to the 1870s, and for decades these forecasts focused on timescales of several hours to days. With today’s high resolution convection-allowing models augmented with machine learning, new research-to-operations tools have emerged from the HWT to bridge that gap between those longer leadtime nationwide forecasts and local severe warnings with leadtimes of only minutes. We’ll highlight a case from the Spring Forecast Experiment where severe storms are rare. Then, review the local Mid-Atlantic July 4th case under the type of operational time constraint faced by participants of the Experiment using the Warn on Forecast System (WoFS). Part 3: Evaluating Deterministic CAMs at the HWT SFE: A Look at the Evaluation Process and GEOS-CAM's Performance Results (Michelle Frazer) A major component of the SFE is daily subjective evaluation of the ability of current state-of-the-art deterministic CAMs to forecast severe convective storms. We'll explore what these daily evaluation activities looked like with examples from this year's SFE. Finally, we'll conclude with presenting the 2025 and 2026 GEOS-CAM objective forecast skill statistics compared to the other participating CAMs, illustrating the model's rapid maturation and its promising future in the landscape of severe weather prediction.