Ronald Errico

Goddard Earth Sciences and Technology Center, University of Maryland, Baltimore County and Global Modeling and Assimilation Office, Goddard Space Flight Center, NASA

The Design, Validation, and Applications of Observing System Simulation Experiments

Room 811 AO&SS, October 5, 2009, 3:30 PM

Abstract

Observing system simulation experiments (OSSEs) are intended to examine the effects of present or proposed atmospheric observing systems on the products of data assimilation systems. Instead of relying on imperfect observations of the imperfectly-known real atmosphere, they employ simulated observations having prescribed errors and drawn from a perfectly-known simulation of nature. By this means, experiments can be performed to estimate otherwise elusive analysis error statistics and the potential of observing instruments yet to be deployed, thus helping to improve data assimilation systems and their products and to better evaluate proposals for competing new instruments. As with any simulation, a good OSSE requires adequate validation of the simulation verses reality as quantified using a variety of metrics. The approach for OSSE development and validation at the GMAO will be described along with some preliminary results. Warnings about how OSSEs can be ill designed and misinterpreted will be offered, along with requirements to ensure that truly useful results be obtained.