Tropical Cyclone Intensity Model Improvement: Better Models or Easier Forecasts?

Tuesday, 19 April 2016
Plaza Grand Ballroom (The Condado Hilton Plaza)
Kieran Bhatia, Univ. of Miami/RSMAS, Miami, FL; and D. Nolan

Handout (1.1 MB)

Over the last 25 years, many studies have asserted that the mean absolute error (MAE) of Atlantic and East Pacific official (OFCL) tropical cyclone (TC) intensity forecasts issued by the National Hurricane Center (NHC) has largely remained constant. However, a recent study by DeMaria et al. (2014) pointed out that the lack of improvement in OFCL intensity forecasts is largely attributable to operational models delivering inconsistent results to NHC forecasters. The authors maintain that intensity guidance has actually improved during this time frame and OFCL forecasts should now follow individual models more closely.

Additionally, NHC verification has highlighted a noticeable drop in the MAE of OFCL forecasts since 2007. This sudden decrease in MAE appears justified. A major model (HWRF) was last added to the suite of intensity guidance for Atlantic and East Pacific basin TCs in 2007, and the best-performing early TC intensity models HWFI, GHMI, LGEM, and DSHP have all surfaced within the last fifteen years. These models have also received numerous upgrades, resulting in an unprecedented period of guidance for forecasters and end users of TC intensity forecasts.

However, the lack of landfalling hurricanes, major hurricanes, and rapid intensification cases since 2007 could also be responsible for the lower MAE of OFCL forecasts. This study investigates whether recent TC seasons were inherently easier for models to forecast or whether model improvement is solely responsible for the lower MAE. To answer this question, a larger sample of verified storms is considered by relaxing some of the verification guidelines followed by NHC and adding the forecasts for storms that dissipate. Model performance is then analyzed for the augmented sample, and verification techniques that account for the difficulty of the forecasts are shared. Unlike DeMaria et al. (2014), where only the trend in MAE is considered, skill score is calculated to account for the forecast difficulty. Then, the frequency of the difficult atmospheric regimes diagnosed by Bhatia and Nolan (2013) and annual average error predicted by PRIME (Bhatia and Nolan 2015) are considered as potential indicators for difficult and easy hurricane seasons for intensity forecasting.

References Bhatia, K. T., and D. S. Nolan, 2013: Relating the Skill of Tropical Cyclone Intensity Forecasts to the Synoptic Environment. Wea. Forecasting, 28, 961–980.

Bhatia, K. T., and D. S. Nolan, 2015: Prediction of Intensity Model Error (PRIME) for Atlantic Basin Tropical Cyclones. Wea. Forecasting, Accepted.

DeMaria, M., C. R. Sampson, J. A. Knaff, and K. D. Musgrave, 2014: Is tropical cyclone intensity guidance improving? Bull. Amer. Meteor. Soc., 95, 387–398.

- Indicates paper has been withdrawn from meeting
- Submission entered in competition