Tests of an ensemble Kalman filter for mesoscale and regional-scale data assimilation

摘要

This dissertation examines the performance of an ensemble Kalman filter (EnKF)implemented in a mesoscale model in increasingly realistic contexts from under a perfectmodel assumption and in the presence of significant model error with syntheticobservations to real-world data assimilation in comparison to the three-dimensionalvariational (3DVar) method via both case study and month-long experiments. The EnKFis shown to be promising for future application in operational data assimilation practice.The EnKF with synthetic observations, which is implemented in the mesoscalemodel MM5, is very effective in keeping the analysis close to the truth under the perfectmodel assumption. The EnKF is most effective in reducing larger-scale errors but lesseffective in reducing errors at smaller, marginally resolvable scales. In the presence ofsignificant model errors from physical parameterization schemes, the EnKF performsreasonably well though sometimes it can be significantly degraded compared to itsperformance under the perfect model assumption. Using a combination of differentphysical parameterization schemes in the ensemble (the so-called ??????multi-scheme?????? ensemble) can significantly improve filter performance due to the resulting betterbackground error covariance and a smaller ensemble bias. The EnKF performsdifferently for different flow regimes possibly due to scale- and flow-dependent errorgrowth dynamics and predictability.Real-data (including soundings, profilers and surface observations) are assimilatedby directly comparing the EnKF and 3DVar and both are implemented in the WeatherResearch and Forecasting model. A case study and month-long experiments show thatthe EnKF is efficient in tracking observations in terms of both prior forecast andposterior analysis. The EnKF performs consistently better than 3DVar for the timeperiod of interest due to the benefit of the EnKF from both using ensemble mean forstate estimation and using a flow-dependent background error covariance. Propercovariance inflation and using a multi-scheme ensemble can significantly improve theEnKF performance. Using a multi-scheme ensemble results in larger improvement inthermodynamic variables than in other variables. The 3DVar system can benefitsubstantially from using a short-term ensemble mean for state estimate. Noticeableimprovement is also achieved in 3DVar by including some flow dependence in itsbackground error covariance.