Although tropical cyclone track forecast errors have substantially decreased in recent decades, there are still cases each season with large uncertainties in the forecasts and/or very large track errors. As such cases are challenging for forecasters, it is important to understand the mechanisms behind the low predictability. For this purpose the research community has developed a number of tools. These tools include ensemble and adjoint sensitivity models, ensemble perturbation experiments and nudging experiments. In this report we discuss definitions of difficult cases for tropical cyclone track forecasts, diagnostic techniques to understand sources of errors, lessons learnt in recent years and recommendations for future work.

This review summarizes techniques used by operational centers to forecast tropical cyclone intensity change as presented to the International Workshop on Tropical Cyclones (IWTC-9) in Hawaii in 2018. Recent advances and major changes over the past four years are presented, with a special focus on forecasting rapid intensity changes. Although intensity change remains one of the most difficult aspects of tropical cyclone forecasting, objective guidance has shown some improvement.
The greatest improvements are realized when consensus methods are utilized, especially those that blend statistical-dynamical based guidance with dynamical ocean-coupled regional models. These models become even more skillful when initialized with inner core observational data. Continued improvement and availability of intensity guidance along with associated forecaster training are expected to deliver forecasting improvements in the future.

This paper summarizes the forecast methods, outputs and skill offered by twelve agencies for seasonal tropical cyclone (TC) activity around the world. These agencies use a variety of techniques ranging from statistical models to dynamical models to predict basinwide activity and regional activity. In addition, several dynamical and hybrid statistical/dynamical models now predict TC track density as well as landfall likelihood. Realtime Atlantic seasonal hurricane forecasts have shown low skill in April, modest skill in June and good skill in August at predicting basinwide TC activity when evaluated over 2003-2018. Real-time western North Pacific seasonal TC forecasts have shown good skill by July for basinwide intense typhoon numbers and the ACE index when evaluated for 2003-2018. Both hindcasts and real-time forecasts have shown skill for other TC basins. A summary of recent research into forecasting TC activity beyond seasonal (e.g., multi-year) timescales is included. Recommendations for future areas of research are also discussed.

Here we discuss recent progress in understanding tropical cyclone (TC) subseasonal variability and its prediction. There has been a concerted effort to understand the sources of predictability at subseasonal time-scales, and this effort has continued to make progress in recent years. Besides the Madden-Julian Oscillation (MJO), other modes of variability affect TCs at these time-scales, in particular various equatorial waves. Additionally, TC activity is also modulated by extratropical processes via Rossby wave breaking. There has also been progress in the ability of models to simulate the MJO and its modulation of TC activity. Community efforts have created multi-model ensemble datasets, which have made it possible to evaluate the forecast skill of the MJO and TCs on subseasonal time-scales in multiple forecasting systems. While there is positive skill in some cases, there is strong dependence on the ensemble system considered, the basin examined, and whether the storms have extratropical influences or not. Furthermore, the definition of skill differs among studies. Forecasting centers are currently issuing subseasonal TC forecasts using various techniques (statistical, statistical-dynamical and dynamical). There is also a strong interest in the private sector for forecasts with 3-4 weeks lead time.    

In order to understand the current and potential use of ensemble forecasts in operational tropical cyclone (TC) forecasting, a questionnaire on the use of dynamic ensembles was conducted at operational TC forecast centers across the world, in association with the World Meteorological Organisation (WMO) High-Impact Weather Project (HIWeather). The results of the survey are presented, and show that ensemble forecasts are used by nearly all respondents, particularly in TC track and genesis forecasting, with several examples of where ensemble forecasts have been pulled through successfully into the operational TC forecasting process. There is still, however, a notable difference between the high proportion of operational TC forecasters who use and value ensemble forecast information, and the slower pull-through into operational forecast warnings and products of the probabilistic guidance and uncertainty information that ensembles can provide. Those areas of research and development that would help TC forecasters to make increased use of ensemble forecast information in the future include improved access to ensemble forecast data, verification and visualizations, the development of hazard and impact-based products, an improvement in the skill of the ensembles (particularly for intensity and structure), and improved guidance on how to use ensembles and optimally combine forecasts from all available models. A change in operational working practices towards using probabilistic information, and providing and communicating dynamic uncertainty information in operational forecasts and warnings, is also recommended.