Sensitivity Analysis in Atmospheric Data Assimilation


Ron Gelaro

NASA Global Modeling and Assimilation Office

Goddard Space Flight Center

Greenbelt, MD 20771


Mathematical adjoints derived from the complex equations comprising a data assimilation and forecast model system have proven to be effective tools for estimating sensitivities in many contexts.  With the adjoint of the analysis component of a data assimilation system, sensitivities of aspects of either forecasts or the analyses themselves can be efficiently estimated.  These can be determined with respect to observational data, background fields or assimilation parameters, all computed simultaneously.  This permits arbitrary aggregation of the sensitivities, e.g., by data type, data channel, data location, etc.  It also allows for estimation of the impacts of any subset of data on standard forecast measures and has proven useful for monitoring observation quality.   The results so far show great promise, but also raise interesting questions about how best to design future strategies for intelligent data selection and utilization.   We’ll examine observation impact results from the adjoint of the Navy’s 3DVAR system (NAVDAS), as well as show preliminary test results from the development of the adjoint of NCEP’s GSI system.