1. INTRODUCTION 1 From a catastrophe modeling standpoint, accurate simulation of tropical cyclone (TC) winds in terms of the maximum wind speed, the distance at which they occur, and how they decay outward (and inward) from the radius of maximum winds are important features for correctly modeling the damage from TCs. Whether those maximum winds occur on the left, right, front, or back of the storm is also an important consideration, given that property along the coast and inland is non-uniformly distributed and valued. Because of the typically hundreds of thousands of events they must simulate, many catastrophe models are inherently parametric in nature and therefore must make simplifying assumptions. For example, they typically assume that TCs have a circularly symmetric rotational wind field and that the forward speed imposes the only horizontal asymmetry, which leads to the strongest winds being on the right-hand side for northern hemisphere storms. However, observational and modeling studies show that wind asymmetries can stem from embedded mesoscale circulations like tornadoes (Schultz and Cecil 2009), environmental wind shear (Frank and Ritchie 1999), adjacent weather systems (Persing et al. 2002), and extratropical transitioning (XTT, Klein et al. 2002, Chen and Yau 2003). The latter is a typical feature of storms affecting Japan. Over 60% and nine of the top ten loss causing events that have affected Japan undergo XTT within 500 km of the coastline (Sousounis and Butke 2010). The transitioning process typically leads to the formation of fronts that can significantly skew the symmetry. In some cases, the evolution leads to the formation of strongest winds on the left hand side, which can conflict with the assumption that strongest winds from a TC are on the right hand side and which can lead to inaccuracies in the modeled damage for a particular storm and also for the overall risk of a particular region. Typhoon Tokage 2004 provides a good example of an asymmetric typhoon that affected Japan. According to the Japanese Meteorological 1* Corresponding author address: Peter J. Sousounis, AIR-Worldwide Corporation., 131 Dartmouth Street, Boston, MA 02116; email: psousounis@air-worldwide.com. Agency (JMA), Tokage made landfall on Shikoku Island at 06 UTC 20 October 2004 with a central sea-level pressure (SLP) of 950 hPa and 75 kt winds (10 min sustained) as shown in Fig. 1. Available observations from the Automated Meteorological Data Acquisition System (AMeDAS) show strongest winds near the center at that time were on the right-hand side (RHS). But, because Tokage was undergoing XTT, its interaction with an upper level jet streak enhanced ageostrophic winds at the surface so that strong winds developed on the left hand side (LHS) – as its central pressure was rising (Sousounis and Desflots 2010). Tokage caused approximately one billion USD in insured loss according to the General Insurance Agency of Japan (GIAJ). A significant portion of it (i.e., 25- 40%) was from precipitation induced flooding and strong winds on the left (west) coast of Japan. It is likely that there are other storms similar to Tokage – in the sense that strong winds develop on the left hand side. But, since not all storms that undergo XTT develop strong winds on the LHS, and since not all storms that develop strong winds on the LHS undergo XTT, it is important to understand some key aspects regarding the climatology of asymmetric TCs so that these features can be incorporated into TC catastrophe models. As a step in that direction, this study examines the climatological frequency with which typhoons affecting Japan exhibit maximum winds on the LHS. It also examines the synoptic scale environment under which a left-hand maximum wind (referred to as Left Max throughout the remainder of this paper) occurs. Both of these features are important to understand so that wind field asymmetries can be incorporated more accurately into the parametric wind field calculation of TC catastrophe models. 2. IMPORTANCE OF ASYMMETRY FOR CATASTROPHE MODELING One may justifiably argue that stochastically modeling the wind damage from storms with random asymmetries is not so significant – if the asymmetries are truly random or very weak. However, the key phrase truly random or very weak is important to consider. If an equal number of storms that affect an area have strongest winds on the left as on the right, then in the aggregate sense, this argument may be 7D.7 TROPICAL CYCLONES RISK MODELS – CIRCULAR SYMMETRY REVISITED Peter J. Sousounis*, Jason Butke, and Kevin Hill AIR Worldwide Corporation, Boston, MA