Observing Strategy and Observation Targeting for Tropical Cyclones using Ensemble-based Sensitivity Analysis and Data Assimilation

A WRF-based Ensemble Kalman Filter (EnKF) data assimilation system, along with the ensemble-based sensitivity analysis, is used to explore observing system design and observation targeting for tropical cyclones. The case we selected is a Western Pacific storm of Typhoon Morakot (2009) which brought record-breaking flooding to the Taiwan Island. The ensemble sensitivity analysis uses the linear correlation between potential observing variables at the targeted time and the forecasting variables at the verifying time evaluated with ensemble perturbations, which has similarity to the adjoint sensitivity analysis. The ensemble sensitivity from a 4.5-km convection-permitting ensemble forecast of 50 members reveals a potential area of high-impact observations located in the southwest quadrant of the typhoon system ~48 prior to its landfall in the Taiwan Island. Numerous observing system simulation experiments (OSSEs) using the WRF-EnKF by assimilating synthetic soundings observations from a reference WRF forecast (“truth simulation”) that closely matched the observed track and precipitation were performed to evaluate the effectiveness of observation targeting using the ensemble sensitivity. While overall there are some positive correlations between the predicted observation impacts using the ensemble sensitivity and the simulated impacts using the EnKF, the effectiveness of such observation targeting strategy appears to be rather limited. The limitations may be due to the strong nonlinearity of the typhoon system, due to the accuracy of the sample covariance from the forecasting ensemble that is based on an imperfect dislocated prior estimate, and/or due to the projection of an individual sounding observation to the complex targeted sensitivity vectors from the ensemble. Further analysis of the OSSE experiments reveals that soundings along the southwest-northeast direction across both the first-guess and simulated truth typhoon centers are the most effective in correcting the uncertainties in the track forecast and subsequent improved precipitation forecast. The generality of this result will also be test for other tropical cyclones. Ongoing OSSE experiment also explores the impact of using different number of soundings in terms of different flight patterns, altitude, and frequency/spacing. The ultimate goal of this research is to improve the observing and targeting strategies of monitoring and forecasting tropical cyclones using the aircraft dropsonde observations.