JAMSTEC > HPCI SPIRE Field3 > Results of Researches > CMSM


A high-resolution atmosphere-ocean coupled model substantially improves typhoon intensity forecasts verified by a large number of experiments using the K computer

Because typhoons are often highly destructive, their accurate prediction is very important in the field of weather forecasting. Although track error in forecasts has been improved during the past two decades, forecast skill of typhoon intensity (minimum sea level pressure and maximum wind speed) is a very challenging topic. Forecasting typhoon intensities is difficult because it depends on fine-scale inner core dynamics, which requires a high-resolution simulation. Moreover, accurate prediction of sea surface temperature beneath a typhoon is also thought to be critical since typhoons can intensify and maintain their circulation by using latent heat release associated with the self-inducement of moist air over warm ocean. Therefore, several researchers speculate that a high-resolution atmosphere-ocean coupled model presumably exhibits the better forecast skill of typhoon intensity. Nevertheless, assessments have never been done before based on a large number of samples partly because such a simulation requires heavy computational resource.

Toward the aim of improving the typhoon intensity prediction, Coupled Meso Scale model (CMSM) has been developed by Japan Agency for Marine-Earth Science and Technology (JAMSTEC), Meteorological Research Institute (MRI), and University of the Ryukyus. CMSM consists of the Japan Meteorological Agency (JMA) Non-Hydrostatic Model (NHM) that can represent the details of the inner core structure of typhoon and an ocean model that accounts for changes in sea surface temperature associated with a typhoon passage. The K computer provides an opportunity of conducting 281 CMSM runs that include all the typhoons approaching mainland Japan from April 2009 until September 2012 (Figure 1). This is the first achievement in the world that quantifies the benefits of a high-resolution atmosphere-ocean coupled model in typhoon intensity forecasts based on a large number of simulations. The results are compared with the atmospheric meso scale model (AMSM) and the JMA global atmospheric spectral model (GSM) that is used for the operational typhoon intensity guidance.

A large number of simulations exhibit that CMSM outperforms the existing forecast models in terms of typhoon intensity forecasts. As for the prediction of minimum sea level pressure, CMSM is better than the other models by about 20-30% at the forecast time of 2 days and by about 30-40% at the forecast time of 3 days. Regarding maximum wind speed, CMSM is better by about 10-20% at the forecast of 2 days and by about 20-30% at the forecast of 3 days (Figure 2). It is because CMSM successfully reproduces time-varying sea surface temperature as consistent with the in situ observations (Figure 3) and can resolve the inner core dynamics of typhoons .

This research clearly demonstrates the benefits of using a high-resolution atmosphere-ocean coupled model that can contribute to disaster prevention and mitigation associated with typhoon-related natural disasters.

The detailed results of this study has been published in Weather and Forecasting, one of the journals of American Meteorological Society, and were presented at the “World Weather Open Science Conference” hosted by World Meteorological Organization in Montreal, Canada.

Ito, K., T. Kuroda, K. Saito, and A. Wada, 2015: A large number of tropical cyclone intensity forecasts around Japan using a coupled high-resolution model. Weather and Forecasting, 30, 793-808.

Figure 1. Calculation domain and verification region. The thin lines indicate 34 typhoons used in this study.

Figure 2. Errors of (a) minimum sea level pressure (MSLP) and (b) maximum wind speed (Vmax). The results are shown for GSM (green), AMSM (red) and CMSM (blue).

Figure 3. Sea surface cooling beneath Typhoon Roki occurred in 2012. Initial time is 21 JST on August 25, 2012.

Corresponding to Kosuke Ito (University of the Ryukyus / MRI)
and Kazuo Saito (MRI / JAMSTEC)
Email: ,