Applying Satellite Observations of Tropical Cyclone Internal Structures to Rapid Intensification Forecast With Machine Learning

Tropical cyclone (TC) intensity change is controlled by both environmental conditions and internal storm processes. We show that TC 24‐hr subsequent intensity change (DV24) is linearly correlated with the departures in satellite observations of inner‐core precipitation, ice water content, and outflow temperature from respective threshold values corresponding to neutral TCs of nearly constant intensity. The threshold values vary linearly with TC intensity. Using machine learning with the inner‐core precipitation and the predictors currently employed at the National Hurricane Center (NHC) for probabilistic rapid intensification (RI) forecast guidance, our model outperforms the NHC operational RI consensus in terms of the Peirce Skill Score for RI in the Atlantic basin during 2009–2014 by 37%, 12%, and 138% for DV24 ≥ 25, 30, and 35 kt, respectively. Our probability of detection is 40%, 60%, and 200% higher than the operational RI consensus, while the false alarm ratio is only 4%, 7%, and 6% higher. Plain Language Summary Improving the predictive skill of TC intensity change, especially rapid intensification (RI), has been a top priority of the National Hurricane Center (NHC). We show TC 24‐hr subsequent intensity change (DV24) is approximately linearly correlated with the departures in satellite observations of inner‐core precipitation, ice water content, and outflow temperature from respective threshold values corresponding to neutral TCs of nearly constant intensity. The threshold values vary linearly with TC intensity. Using machine learning techniques, we construct a statistical model for RI forecast with a combination of the predictors included in the NHC probabilistic forecast guidance and satellite observations of storm internal structure. The overall predictive skill of our model for 2009–2014 RI in the Atlantic basin is better than the NHC operational RI consensus by 37%, 12%, and 138% for DV24 ≥ 25, 30, and 35 kt, respectively. The probability of detection for RI by our model is 40%, 60%, and 200% higher than the operational RI consensus, while the false alarm ratio is only 4%, 7%, and 6% higher. Key Points Tropical cyclone intensity change is approximately linearly correlated with surplus inner‐core precipitation above a threshold value The threshold value can be determined from neutral TCs of nearly constant intensity, and it increases linearly with TC intensity Machine learning with combined environmental factors and surplus precipitation signif