Ensemble coupling assimilation system and method for numerical forecasting

Abstract

The invention discloses an ensemble coupling assimilation system and method for numerical forecasting. The ensemble coupling assimilation system comprises a coupling mode integration and cooperation set operation management module, an assimilation algorithm integration module, a set coupling assimilation test configuration module and a set coupling assimilation online interaction module. Cooperative operation of all set members under the same MPI task in the same coupling mode can be realized. Different assimilation algorithm instances are flexibly used in different component modes, one assimilation algorithm instance is supported to use all MPI processes of all set members in the corresponding component mode for parallel operation, and finally the operation efficiency of a corresponding assimilation system and the flexibility of assimilation configuration are improved.

Description

technical field [0001] The invention relates to the technical field of numerical prediction. More specifically, the present invention relates to an ensemble coupled assimilation system and method for numerical prediction. Background technique [0002] In response to the demand for extended forecast timeliness and seamless forecast business development, the traditional atmospheric model-based numerical weather prediction is developing towards the direction of coupled model and coupled assimilation technology. Compared with the coupled model, coupled assimilation is in the initial stage of development. Not only the assimilation method faces many challenges, but also how to build a general coupled assimilation framework is a major challenge in its intersection with computer science. [0003] The four-dimensional variational method based on optimal control and the ensemble Kalman filter based on ensemble forecasting are advanced assimilation methods that are widely used in busi...

AI Technical Summary

Problems solved by technology

Although PDAF integrates multiple ensemble assimilation algorithms and has been applied to the development of assimilation systems for multiple modes, its online interaction method is difficult to support coupled modes, that is, it is difficult to support coupled assimilation, because: 1) PDAF requires assimilated component modes All MPI processes are used, and in most of the coupled modes, a component mode only uses part of the MPI processes of the coupled mode; 2) In coupled assimilation experiments, it may be necessary to use different assimilation algorithms to assimilate multiple component modes, but PDAF allows only one assimilation algorithm, which imposes limitations on coupled assimilation

In addition, PDAF requires the assimilation algorithm to only use the MPI process of the first set member of the pattern, which brings a limit to the speed of the assimilation algorithm

[0006] To sum up, the current development of set coupling assimilation is facing severe technical challenges. There is no set coupling assimilation framework based on efficient online interaction at home and abroad, and only low-efficiency offline interaction based on file reading and writing can be used to develop set coupling. assimilation system

Images