Regional and sub-basin seasonal tropical cyclone activity in the CMCC-SPS3.5 model

Reliable seasonal forecasts of tropical cyclone (TC) activity are fundamental in helping stakeholders make informed decisions and mitigate economic and societal losses. While several public and private institutions issue seasonal forecasts of tropical storms for traditionally investigated basins, like the North Atlantic and the Western North Pacific, only a few provide global coverage, limiting confidence for other densely inhabited regions. Here, we evaluate the retrospective seasonal forecasts of TC activity across five basins (North Atlantic, Eastern and Western North Pacific, South Indian and South Pacific) over the period 1993-2016, using the Euro-Mediterranean Center on Climate Change Seasonal Prediction System 3.5 (CMCC-SPS3.5), a coupled general circulation model used for operational seasonal forecasts. CMCC-SPS3.5 skillfully captures key features of TC climatology (i.e., spatial distribution and seasonal cycle) and predicts with statistically significant skill their interannual variability, both in terms of numbers of tropical cyclones (NTC) and pressure-based accumulated cyclone energy (PACE). The model shows asymmetric performance, with TC activity overestimated in the Southern Hemisphere and underestimated in the Northern Hemisphere compared to the observations. Using a probabilistic clustering approach, we show that the model has statistically significant skill in year-to-year variability for specific track patterns across each basin. Using the North Atlantic basin as a case study, we show that the ENSO-TC teleconnection is stronger in CMCC-SPS3.5 compared to the observations, with implications for cyclone predictability. Moreover, our findings suggest that the basin-wide predictability is the result of the cumulative skill of individual clusters, providing insights for developing track-based forecasts. This study also demonstrates the readiness of CMCC-SPS3.5 for operational global TC seasonal forecasting.