Determination of the Effects of Non-Modality on the Data Assimilation and Numerical Weather Prediction Problems

University Of Maryland Baltimore County, Baltimore MD

Investigators

Abstract

The PIs will address the non-modal growth of forecast errors in numerical weather prediction. Errors after day one into the forecast have large vertical and horizontal scales, but their incipient cause has been found to be due to errors in much smaller spatial scales and to have different structures such as being tilted and dynamically unbalanced as opposed to barotropic and geostrophic. Such error behavior is called non-modal. Based on singular vector analysis, the implications of non-modal forecast error growth for potential improvement of both weather forecasting and analysis of observational data will be determined, and forecast sensitivity with respect to observational data errors will be examined in order to better understand the impacts of the observing system imperfections on forecast and data assimilation systems. These questions will be addressed by applying the adjoint version of the data assimilation and weather modeling system recently developed at NASA's Global Modeling and Assimilation Office. This research has the potential to improve weather forecasts which are important for societal activities.