[p1] Multi-scale and multi-component treatments


A case study involving single observation experiments performed over snowy Siberia using a coupled atmosphere-land modeling system

Kazuyoshi Suzuki (JAMSTEC), Milija Zupanski (Colorado State University), Dusanka Zupanski (Spire Global Inc.)


Through a series of single observation experiments, we investigate forecast error covariance and correlation structures by applying the Maximum Likelihood Ensemble Filter (MLEF) data assimilation method with a coupled atmosphere-land surface model. We consider a coupled approach that includes the full complexity of the cross-component covariance and correlation structures, specifically between 2-m air temperature and snow temperature. We show that the error covariance and correlation are complex and are strongly dependent on both weather conditions and the land surface scheme, indicating a need for further investigation of the relevance of flow-dependent ensemble covariance in coupled systems. We also demonstrate that the use of coupled error covariance methods improves the efficiency of information transfer between the atmosphere and the land surface by allowing the well-observed atmosphere to influence land surface variables.