Tropical Cyclone Forecast Verification: Multivariate Attributes and Rapid Intensity Change

Tropical cyclones (TC) are a serious concern for the nation, causing significant risk to life, property and economic vitality. In recent decades, TC forecast position errors have been reduced substantially through improved model guidance, enhanced observations, and forecaster expertise. However, little progress has been made during this period in reducing TC forecast intensity errors. In response, NOAA established the Hurricane Forecast Improvement Project (HFIP) in 2009. HFIP provides the basis for NOAA and other agencies to coordinate hurricane research needed to significantly improve numerical guidance for TC forecasts, with a focus on rapid intensity change since models have not reliably simulated this type of event.

Evaluating TC forecast position and intensity errors independently has been a common verification practice. However, TC environments can significantly impact intensity (ex. warm versus cold sea surface temperature, TC interaction with land, etc.). This paper will present results of an exploratory multivariate analysis wherein conditional distributions and characteristics of TC forecast position and intensity errors will be considered. The TC forecast data sets were provided by HFIP model development groups as part of annual retrospective testing and demonstration exercises. Additionally, this paper will examine the frequency of TC rapid intensity change events in the National Hurricane Center Best Track Analyses and suggest frequency-appropriate statistics for these events.