An adjoint modeling system based on the Naval Research Laboratory's (NRL) Coupled Ocean-Atmosphere Mesoscale Prediction System (COAMPS) atmospheric model was developed. The tangent linear and adjoint codes for the portions of the model pertaining to the dynamical core and the planetary boundary layer processes were formulated and tested. The system was structured to be used on multiple processors of a distributed memory system, as was the version of the forward model on which the adjoint modeling system is based.

The COAMPS adjoint model allows for the COAMPS model to be used in a four-dimensional variational (4D-Var) data assimilation scheme. A Bogus Data Assimilation (BDA) scheme was implemented with the COAMPS atmospheric model. Results from forecasts of tropical cyclones with the COAMPS modeling system will be compared to forecasts produced by the fifth generation Penn State / National Center for Atmospheric Research Mesoscale Model (MM5) using the same hurricane initializaiton scheme.

The tangent linear and adjoint codes are currently being developed for the newest version of the explicit moisture scheme developed for the COAMPS model. These codes will be added to the adjoint modeling system to allow data assimilation to be conducted at cloud resolvable scales and microwave radiance obsverations to be assimilated in tropical cyclones, both of which will hopefully lead to improved hurricane forecasting.