Radar Data Assimilation Approach For The Forecasting Of Landfalling Typhoon

This study explores the approach for radar data assimilation (DA) and its application in tropical cyclone (TC) forecasting. From the perspective of improving the quality of typhoon analysis and subsequent forecast, the assimilation of T-TREC (Typhoon circulation-Tracking Radar Echo by Correlation) winds retrieved from single Doppler radar radial velocity (Vr) and reflectivity (Z) data are investigated as compared to the direct assimilation of Vr data, for two landfalling typhoons, Meranti (2010) and Chanthu (2010). The potential advantage of T-TREC assimilation for real time forecasting is also illustrated. Moreover, an enhanced hybrid DA scheme is proposed with dynamic constraint included in the hybrid DA framework. The dynamic-constrainted hybrid Ensemble-3DVAR scheme combines, for the first time, the dynamic constraint and flow-dependent covariance in the variational (VAR) DA framework. Based on the newly developed assimilation scheme, as viewed from a better forecasting quality and a longer warning time, the performance of cycled DA for T-TREC winds and Vr data is investigated, respectively. Besides, the combined DA strategy is also proposed assimilating two kinds of radar wind data in different cycles. The conclusions are summarized as follows:1. An extended Tracking Radar Echo by Correlation (TREC) technique, called T-TREC technique, has been developed recently to retrieve horizontal circulations within TCs from single Doppler radar reflectivity (Z) and radial velocity (Vr, when available) data. This study explores, for the first time, the assimilation of T-TREC-retrieved winds for a landfalling typhoon, Meranti (2010), into a convection-resolving model, the WRF (Weather Research and Forecasting). The T-TREC winds or the original Vr data from a single coastal Doppler radar are assimilated at the single time using the WRF 3DVAR, at 8,6,4 and 2 hours before the landfall of typhoon Meranti. In general, assimilating T-TREC winds results in better structure and intensity analysis of Meranti than directly assimilating Vr data. The subsequent forecasts for the track, intensity, structure and precipitation are also better, although the differences becomes smaller as the Vr data coverage improves when the typhoon gets closer to the radar. The ability of the T-TREC retrieval in capturing more accurate and complete vortex circulations in the inner-core region of TC is believed to be the primary reason for its superior performance over direct assimilation of Vr data; for the latter, the data coverage is much smaller when the TC is far away and the cross-beam wind component is difficult to analyze accurately with 3DVAR method.2. Single observation tests are used to investigate the advantage of hybrid DA method as compared to 3DVAR method. For typhoon chanthu (2010), the analysis increments from 3DVAR method which only uses static covariance reveals no specific structure related to tropical cyclone. While the hybrid DA method which employs full weight ensemble covariance produces more realistic vortex structure with flow-dependent characteristic. When the hybrid DA includes both static covariance and ensemble covariance, the incremental structure is actually in between. For the adjustment of typhoon structure, the hybrid method with full weight ensemble covariance shows the best performance. Besides, the ensemble covariance localization is also examined for hybrid DA. Based on the sensitivity tests,50 km (4 km) of horizontal (vertical) localization radius are suitable for radar data assimilation.3. A dynamic equation based on the steady momentum equations is incorporated into the WRF hybrid Ensemble-3DVAR (En3DVAR) DA system as a weak constraint to form the dynamic-constrainted En3DVAR scheme. This study explores for the first time the combination of both ensemble flow-dependent covariance and dynamic constraint in VAR cost function for the assimilation of TCs. The goal of the newly developed scheme is to build a more flow-dependent analysis with dynamic consistency with the use of radar wind data when only the momentum fields are assimilated. The scheme is applied to a landfalling typhoon, named Chanthu (2010), at a convection-allowing resolutions assimilating radar T-TREC retrieved winds with complete TC circulation provided in high retrieval quality. Response to the significant cyclonic wind increments brought from the assimilation of T-TREC winds, the flow-dependent covariance helps to update unobserved mass field in a more dynamically and thermodynamically way compared to static covariance, while the imposed dynamic constraint is effective in further deepening the surface pressure. The better conditioned pressure field in dynamic-constraint En3DVAR also adjusts the temperature and moisture within TC inner core through the impact of flow-dependent covariance. The deterministic forecasting results show that compared to traditional 3DVAR and En3DVAR, the newly proposed dynamic-constrainted En3DVAR produces the best intensity forecasts especially in terms of minimum sea level pressure. Sensitivity experiments concerning with the chosen of different dynamic weight in the cost function are also examined.4. The dynamic-constrainted hybrid En3DVAR scheme is employed to investigate the performance of cycled DA for Vr data and T-TREC winds, respectively. Moreover, considering the respective advantage for T-TREC winds and Vr observation from multiple radars, the combined DA strategy is proposed in this study assimilating two kinds of radar wind in different cycles. The cycled assimilation of Vr data is effective in improving the typhoon analysis and forecast. However, the good performance requires numerous DA cycles, leading to a relative latter starting time of deterministic forecast which is not favorable for real time. On the contrary, the cycled assimilation of T-TREC winds significant improves the typhoon intensity forecast after only a few cycles, benefiting from its advantage of larger coverage and more complete wind circulation. However, when the number of DA cycles increase, the degraded forecasting performance is attributed to the accumulation of T-TREC retrieval errors during the analysis cycles. For the newly proposed combined DA strategy, the deterministic forecasts show the best results among three DA strategies employing each launch time.