This startup was founded in 2024 by Chanh Kieu, an Associate Professor in the Department of Earth and Atmospheric Sciences at Indiana University Bloomington. It is dedicated exclusively to predicting tropical cyclone (TC) intensity through advanced Artificial Intelligence (AI) technology, reflecting our commitment to improve environmental risk management for society via technology. By introducing an innovative approach to TC intensity forecasting and harnessing state-of-the-art AI models, we wish to deliver a cutting-edge consulting solution to different sectors with greater efficiency, flexibility, and reliability.

The startup began as a side project on the predictability of TC rapid intensification at Indiana University Bloomington during May-August 2021. It quickly evolved into exploring the use of AI for operational TC intensity forecasting. Our in-house TC intensity forecast products were first generated in real-time during the summer 2024 hurricane season. This real-time experiment was then followed by the launch of our official website, tchunting.net, in October 2024, enabling us to broadcast our products live, with additional support for seasonal outlooks and climate projections.

Unlike the traditional regional TC prediction models, our regional AI-based models for TC intensity prediction are built purely on a data-driven framework, with some enhancement from the TC-scale dynamics and time stepping techniques for spatiotemporal problems in deep learning. Key features of our regional AI-based TC models include:

• The ability to provide intensity predictions for any lead time however long it is, with or without boundary conditions from global models;
• Flexibility to predict TC intensity along prescribed tracks or across track scenarios;
• The ability to quickly incorporate TC intensity fluctuations for any nearby tracks;
• High efficiency and adaptability to a wide range of global inputs from various climate projections and scenarios;
• Rapid re-training or adaptation to upgrades in global models such as GFS, ECMWF, or others;
• Expandability to integrate with new TC-environment prediction system or other global AI models.

For the current products, our 1-7 day intensity forecast takes input data from the NOAA/NCEP global GFS model and the regional HFSA model to generate initial and boundary conditions for our AI-based models. For 3-6 month outlook products, the global NCEP CFS real-time forecast is used, while 5-20 year TC projections are based on the most recent CMIP6 multi-model products. All of these datasets are open to public and can be accessed directly on their data portals. Both the back-end AI models and front-end workflows are designed to ensure seamless, real-time updates of TC intensity forecasts.

These real-time TC intensity prediction products represent the culmination of our nearly 25 years of research on TC dynamics and modeling. However, as always with any forecast or projection, there are some inherent uncertainties in all of our products. As such, we provide these products as an additional source of information for consultation, but users are solely responsible for their decisions. We will assume no liability for outcomes resulting from the use of these products.

There are many further features that we plan to include and/or improve in the future. Any comments or suggestions, please email us here.