Tropical Cyclogenesis Hypothesis Testing Using Ensemble Sensitivity Analysis

Ensemble sensitivity analysis is used to test the sensitivity of developing tropical cyclones to preceding conditions. Ensemble sensitivity analysis exposes the spatial and temporal relationships between field variables, which can then be used to systematically perturb states preceding development to test the key factors that control development. Ensemble samples are derived from an ensemble Kalman filter (EnKF), which is cycled on conventional and special field observations gathered during the TPARC/TCS08 campaign. Four storms are considered: Nuri, Sinlaku, Hagupit and Jangmi. A variety of metrics are defined to determine the sensitivity of the developing cyclone to states 24-hours prior, including low-level and mid-level circulation, and measures of water vapor. Initial-value experiments are devised based on these sensitivity results to "control" the evolution of the typhoons over a 24 hour forecast cycle. For example, moisture-denial experiments are performed ,where all degrees of freedom except the moisture fields are perturbed in the initial conditions to isolate the contribution from dry and moist dynamics on the storm development. Similar experiments involve altering the mid and low level circulation to test hypotheses regarding "top-down" and "bottom-up" development. Results consistently show, across all four storms, that the most important aspect preceding development is circulation near 850 hPa. Moreover, sensitivity to dry dynamics strongly dominates the contribution from moisture, except in the case of Nuri. The range of validity of these results is sufficiently broad to allow the storms to be completely eliminated, or approximately doubled in intensity, using the sensitivity fields.