Since tropical cyclones (TCs) are often highly destructive, the improvement in the track and intensity forecasts is important in terms of disaster prevention and mitigation. Generally speaking, a TC track is primarily determined by the large-scale environment in which the tropical cyclone is embedded, while the TC intensity is dependent on the smaller scale internal dynamics as well as its interaction with the environment. Thus, we should select a better metric and control variables suited to each problem toward the improved forecasts. In this presentation, I will talk on several works that I and collaborators have recently engaged in.
First, I introduce a new sensitivity analysis in which a TC-position itself is taken as a metric. It objectively identifies the important regions toward the reduction in the track forecast errors. This sensitivity is interpreted as the slope of regression line (or its relative) between the TC position and initial perturbation based on ensemble run by use of an incremental approach. Besides its clear objectivity, this Incremental Typhoon-position-Oriented Sensitivity Analysis (ITO-SAn) has an advantage of extended applicability to the cases with a high nonlinearity even when the distribution of ensemble TC positions has bimodal peaks since the linear time evolution of perturbations is not assumed. As a first step toward illustrating ITO-SAn's clear objectivity and versatility, it is applied to the case of TC Shanshan (2006). The sensitivity field of the typhoon central latitude with respect to vorticity field is characterized by several features such as the swirling pair pattern centered at the initial typhoon position and the signature of Rossby waves away from the TC center.
As for the intensity modeling, it is well known that uncertainty in the values of air-sea heat and momentum exchange coefficients has a detrimental effect since TCs intensify and maintain their circulations against surface friction through the self-inducement of anomalous heat fluxes from the sea surface. Thus, we have proposed the optimization of the air-sea exchange coefficients (in addition to the initial value) through the variational data assimilation (DA) method toward the better TC intensity representation. In an OSSE experiment with the idealized TC model, these coefficients are successfully improved toward the true value by digesting the available observational data, and it yields the improvements in the analysis field of the maximum wind speed and the inner core structure.