Sensitivity Analysis of the Sequential Data Assimilation Methods
Han Pe
Abstract
Han Pe
Abstract
In order to explore the influence of different assimilation indicators on the sequential data assimilation,in this paper, based on Lorenz-1963 model, we attempt to test the sensitivities of three typical assimilation methods, including En KF, DEn KF and En SRF. We have investigated the effects of different indicators on the data assimilation results. These indicators include the total assimilation time, the assimilation step, the ensemble number, the inflation factor, the observation number and the localization radius. The experimental results show that the total assimilation time, the assimilation step, the ensemble number, the inflation factor, the observation number and the localization radius directly affect the data assimilation result. The improved algorithm based on En KF is superior to the En KF under a given condition. When some preconditions are satisfied, the data assimilation results of all above methods tend to be the same. In practice, this work is of great significance in choosing the optimal sequential assimilation method.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
In order to explore the influence of different assimilation indicators on the sequential data assimilation,in this paper, based on Lorenz-1963 model, we attempt to test the sensitivities of three typical assimilation methods, including En KF, DEn KF and En SRF. We have investigated the effects of different indicators on the data assimilation results. These indicators include the total assimilation time, the assimilation step, the ensemble number, the inflation factor, the observation number and the localization radius. The experimental results show that the total assimilation time, the assimilation step, the ensemble number, the inflation factor, the observation number and the localization radius directly affect the data assimilation result. The improved algorithm based on En KF is superior to the En KF under a given condition. When some preconditions are satisfied, the data assimilation results of all above methods tend to be the same. In practice, this work is of great significance in choosing the optimal sequential assimilation method.
Key concepts: Data assimilation, Assimilation (phonology), Mathematics, Statistics, Computer science, Econometrics, Algorithm, Meteorology