Towards a Self-Contained Analysis and Reanalysis System for Hurricanes: A basin-scale ensemble Data Assimilation (DA) framework in the Hurricane Analysis and Forecast System (HAFS)
Joey Knisely
U. of Maryland
6 February, 2026, 12:30pm
Abstract:
Tropical cyclone (TC) prediction is one of the most challenging problems in atmospheric science. While track forecasts have improved, intensity forecasting lags behind. Modern systems like NOAA's HAFS address some challenges but rely heavily on global models for initial conditions and bias correction, limiting the ability to diagnose internal model deficiencies.
This research advances hurricane forecasting through three complementary studies, culminating in the development of a self-contained, basin-scale ensemble data assimilation and forecasting system. The study demonstrates that HAFS can operate in a fully-cycled mode with online satellite radiance bias correction. Experiments show that training bias correction coefficients on the native HAFS grid improves forecasts compared to using external global models. Furthermore, comparing different resolution configurations reveals substantial differences in cloud coverage and composition, generating significant domain-wide biases with consequent impacts on TC forecasts.