296 High resolution precipitation forecast using WRF model during super storm Ida over the New York City metropolitan area

Tuesday, 30 January 2024
Hall E (The Baltimore Convention Center)
Jorge Humberto Bravo Mendez, Stevens Institute of Technology, Hoboken, NJ; and M. Temimi and M. Abdelkader

The main goal of this research is the assessment of the performance of operational regional weather forecasting using the Weather and Forecast Research (WRF) model, version 4.4.2. Employing a three-tiered nested domain approach with resolutions of 9km, 3km, and 1km, we effectively downscaled the Global Forecast System (GFS) forecast during super storm Ida over the NYC metropolitan area. WRF was initialized using GFS forecasts between 2021-08-30 and 2021-09-01 in six-hour intervals, resulting in twelve distinct GFS outputs over a span of 120 forecast hours.

After transitioning from a tropical system, Ida evolved into an extra-tropical system, influenced significantly by interactions with an Atlantic frontal system. These dynamics sustained Ida's convective potential, resulting in substantial precipitation events over the Northeastern US. Utilizing the WRF model, our goal is to reproduce the atmospheric conditions fostering this convective behavior from historical forecast data. To ensure a robust comparative analysis, historical station records were sourced from three distinct repositories: the National Centers for Environmental Information (NCEI) of NOAA (24 stations), Mesonet NY (18 stations), and Mesonet NJ (35 stations).

Seven statistical metrics were employed to critically assess the models' outputs against on-the-ground measurements. These metrics encompassed parameters such as the difference between precipitation peaks, coefficient of determination, Precent Bias and the Kling-Gupta Efficiency. To enhance the spatial validation of the WRF simulations, cumulative rainfall maps were juxtaposed against Multi-Radar/Multi-Sensor System (MRMS) data. Specifically, WRF showcased enhanced KGE values and reduced PB values across varying forecast lead times. A key observation was the temporal disparity in precipitation peaks between WRF and GFS. As the event's peak approached, WRF exhibited markedly superior performance relative to GFS predictions.

Overall, the research underscored WRF's potential to substantially augment early warning systems and drive regional flood models which makes it a candidate for an integration in the operational Stevens Flood Advisory System that uses meteorology from global models. This work sets a precedent for blending Mesonet observations from two distinct networks for a comprehensive assessment of weather forecast throughout the NYC metropolitan area, promising enhanced stakeholder engagement and enriched weather and climate communication. Future work will focus on assessment of the model’s sensitivity to microphysics and PBL schemes and their impact on precipitation forecast.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner