Landfall forecast biases for Gulf of Mexico tropical cyclones

Once a tropical cyclone develops in, or enters, the Gulf of Mexico critical decisions such as: evacuation, resource/aid pre-placement and petrochemical plant shutdowns, have to be made by various interests. Recently, based on the experience from major Hurricanes Katrina and Rita, decisions are being made much earlier than with previous expected landfalls.

For example, during the threat of Hurricane Dean (2007), evacuation timelines and resource/aid pre-placement decisions were made by the State of Texas based on the National Hurricane Center's (NHC) 120 hour forecast of tropical storm force winds reaching the Texas coastline. Given the magnitude of forecast track (both temporal and spatial) and intensity errors typical for this long lead time, it was critical that decision makers -- and the meteorologists who were briefing them -- were aware of any historical forecast biases that might be identified. The goal of this study was to aid the decision making process by examining recent Gulf of Mexico landfall forecasts and noting any substantial forecast biases in either landfall location or intensity at landfall.

Tropical cyclone best track data and NHC forecast advisories and discussions were analyzed for the period of 1998 to 2007. During this period 48 tropical cyclones made landfall along the Gulf of Mexico. Once landfall was predicted within the forecast period (72 hours for 1998-2002), (120 hours for 2003-2007) for any tropical cyclone, the data were analyzed and landfall forecast biases were calculated. The biases were grouped based on the tropical cyclone's initial conditions such as intensity, speed of movement, location, etc., at the time the forecast was issued. Results included a substantial left-of-track, or westward, bias for both tropical storms and hurricanes during most landfall forecast periods, and a later than actual landfall bias for slow moving tropical storms and hurricanes.