Center Fixing of Tropical Cyclones Using Uncertainty-Aware Deep Learning Applied to High-Temporal-Resolution Geostationary Satellite Imagery

Abstract Determining the location of a tropical cyclone’s (TC) surface circulation center—“center fixing”—is a critical first step in the TC-forecasting process, affecting current/future estimates of track, intensity, and structure. Despite a recent increase in automated center-fixing methods, only one such method [Automated Rotational Center Hurricane Eye Retrieval, version 2 (ARCHER-2)] is operational, and its best performance is achieved when using microwave or scatterometer data, which are often unavailable. We develop a deep learning algorithm called GeoCenter; besides a few scalars in the operational Automated Tropical Cyclone Forecasting System, it relies only on geostationary infrared (IR) satellite imagery, which is available for all TC basins at high frequency (10 min) and low latency (<10 min) during both day and night. GeoCenter ingests an animation (time series) of IR images, including nine channels at lag times up to 4 h. The animation is centered at a “first guess” location, offset from the true TC-center location by 48 km on average and sometimes >100 km; GeoCenter is tasked with correcting this offset. On an independent testing dataset, GeoCenter achieves a mean/median/root-mean-square (RMS) error of 26.6/22.2/32.4 km for all systems, 24.7/20.8/30.0 km for tropical systems, and 14.6/12.5/17.3 km for category 2–5 hurricanes, respectively. These values are similar to ARCHER-2 errors with microwave or scatterometer data and better than ARCHER-2 errors when only IR data are available. GeoCenter also performs skillful uncertainty quantification, producing a well-calibrated ensemble of 150 TC-center locations. Furthermore, all predictors used by GeoCenter are available in real time, which would make GeoCenter easy to implement operationally every 10 min. Significance Statement Estimating the location of a tropical cyclone’s (TC) surface circulation center is a critical first step in the TC-forecasting process. Current and future estimates of several TC properties—including the TC track, intensity, and structure—are highly sensitive to this initial location estimate, called the “center fix.” This paper describes a new deep learning algorithm for center fixing, called GeoCenter, whose main input is an animated time series of infrared (IR) satellite imagery. GeoCenter performs competitively with existing methods for center fixing, both operational and nonoperational, and provides skillful estimates of uncertainty in the TC-center location. Furthermore, GeoCenter is designed so that it could be easily implemented in operations.