Wednesday, 13 January 2016
With the adjoint of a data assimilation system, the impact of any or all assimilated observations on a measure of short-range forecast skill can be estimated accurately and efficiently. The approach is especially well suited for assessing the impact of hyper-spectral satellite instruments on numerical weather forecasts because it easily allows aggregation of results in terms of individual data types, channels or locations, all computed simultaneously based on a single pass of the adjoint system. The Global Modeling and Assimilation Office (GMAO) has been computing these impacts routinely since 2010 as part of its Goddard Earth Observing System (GEOS-5) real-time production suite. In this study we examine observation impacts in GEOS-5 24-hour forecasts in light of recent system upgrades – including the implementation of an ensemble-variational hybrid assimilation algorithm and incorporation observations from Suomi-NPP, MetOp-A/B and other sensors – as well as upgrades to the observation impact calculation itself – including the incorporation of moist convection and cloud microphysical processes in the GEOS-5 adjoint model. It is found that both conventional and satellite observations contribute significantly to the reduction of forecast errors, with asymmetries in the magnitudes of their impacts depending on the season and hemisphere. Microwave sounders still provide the largest combined impact overall, but the combined impact of hyperspectral IR sounders has become nearly as large. These impacts can vary significantly from sensor to sensor and channel to channel, with some channels, particularly those most sensitive to water vapor, providing mixed beneficial and non-beneficial impacts in an average sense.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner