Two methods of reducing errors introduced by rapidly changing conditions are considered in this study: 1) Hybrid 4DEnVar, and 2) Hybrid 3DEnVar with hourly frequency (hourly-3DEnVar). In 4DEnVar, the temporal evolution of error covariances is incorporated by the use of 4-dimensional ensemble perturbations with implicit linearity assumption during the minimizations. However, non-linear error growth within a 6-hour window can still pose difficulties. Hourly-3DEnVar assimilates observations in one hour windows instead of one longer six hour window, allowing the error covariances to change each hour. However, frequent interruption of the model could introduce additional instabilities. While previous work has been done to evaluate different data assimilation frequencies and methods on thunderstorms, little work has been done comparing these two methods in the case of a hurricane. The goal of this study is to evaluate the performance of the hourly-3DEnVar and 4DEnVar hybrid data assimilation systems in the HWRF model assimilating inner core Tail Doppler Radar (TDR) data to understand the benefits of each method in hurricane forecasting.
Experiments are conducted for cycles in which TDR data is available during Hurricane Edouard (2014). Edouard became a tropical storm on Sep. 12 and remained tropical until Sep. 19, reaching a peak intensity of 105 knots on Sep 16, becoming a category 3 hurricane. TDR data is available from 4 flights over 5 DA cycles from September 15-17.
A control run is performed using 6-hourly Hybrid-3DEnVar. 4DEnVar and hourly-3DEnVar experiments are run for each DA cycle where TDR data is available. All experiments started from the same the background produced by the control run. To reveal the impact of relocation, hourly-3DEnVar and 4DEnVar were run with and without vortex relocation.
It was found that the both hourly-3DEnVar and 4DEnVar performed better than 6-hourly 3DEnVar when verified against best track intensity and position regardless of whether vortex relocation was performed. This result indicates the importance of accounting for the rapid evolution of the background error covariance. It was also found that both 4DEnVar and hourly 3DEnVar produced forecasts that verified better against best track when vortex relocation is employed. Additionally, while greater imbalance were found in hourly-3DEnVar than those of other experiments, instability propagates outward rapidly and does not result in a significant degradation of the analysis and forecast. Further analyses are being conducted on additional hurricanes, which will be discussed in the conference.