Asymmetry Metrics to Describe the Distribution of Tropical Cyclone-Related Hazards

Knowledge of the spatial distribution of hazards associated with a landfalling tropical cyclone (TC) is necessary for understanding and communicating the risks that a storm poses to individuals and society. Many TCs, even those that landfall as intense hurricanes, have substantial track-relative asymmetries in the wind, surge, and precipitation threats that they present. Although Harvey and Irma both made landfall in the United States as major hurricanes, each storm manifested asymmetric wind, rain, and storm surge hazards. Harvey produced catastrophic freshwater flooding east of its center due to slow motion and a persistent onshore flow, while dry air and an offshore flow greatly limited precipitation west of its center. Irma generated substantial impacts far from its center on its eastern side (e.g. storm surge flooding in Jacksonville and Charleston), while those on its western side were much milder, even at locations close to the center (e.g. Tampa).

The impacts of Harvey and Irma, along with the asymmetric impacts of Irene (2011) and Sandy (2012) indicate the need for objective measures of TC hazard asymmetry. This is especially true for TCs that make landfall while undergoing extratropical transition, a phase during which the wind and precipitation asymmetries associated with the cyclone often change rapidly. Current cyclone structure indices (e.g. Cyclone Phase Space; Hart 2003; Hart and Evans 2003) are useful for diagnosing the thermodynamic structure of a cyclone, but do not provide information on the distribution of potential hazards associated with a storm. Here, we create quantifiable metrics of TC wind and precipitation asymmetry and apply these metrics to the lifecycle of recent high-impact Atlantic hurricanes and cases of extratropical transition.

To calculate the wind hazard asymmetry metric, the square of the 1-minute, 10-meter sustained wind speed is integrated on the right and left semicircles of the TC. The limit of integration is set to the outer radius of 17.5 m s^-1 (39 mph) wind speed in the TC circulation. The difference between the integrals on the right and left side of the circulation is then divided by one half of the integrated value throughout the TC to yield a normalized wind hazard asymmetry value. For TCs near land, land-related friction will reduce 10-meter wind speeds in the portion of the TC over land although the weaker wind speeds are not representative of the storm-scale lower-tropospheric circulation weakening. Thus, land interaction can produce undesirable artificial changes in the wind hazard asymmetry metric. Therefore, we also calculate this metric using wind speeds at 925 and 850 hPa and compare these results to those calculated using 10-meter winds.

The precipitation hazard asymmetry metric is calculated similarly. Using the outer radius from the wind asymmetry metric, the mean precipitation rate on the right and left semicircles of the TC is calculated. This difference is divided by the mean precipitation rate throughout the TC to yield a normalized precipitation asymmetry value. The TC wind and precipitation asymmetry values are represented in two-dimensional phase space, similar to the representation of two of the parameters in CPS (e.g. B vs. -V^T_L). The TC’s path through the two-dimensional phase space represents the time-evolution of wind and precipitation asymmetry metrics.

We examine the evolution of TC wind and precipitation hazard asymmetry for the lifecycle of Harvey, Irma, Irene, and Sandy, as well as several recent cases of Atlantic extratropical transition. Wind and precipitation asymmetry values are calculated using analysis and short-range forecast data from the European Centre for Medium-Range Weather Forecasts (ECMWF) operational forecast and the Global Forecast System (GFS) FNL operational data. We show how each analysis represents the evolution of the TC hazard asymmetries, demonstrating the usefulness of this metric for understanding the real-time and forecast hazard distributions of future TCs.