General Description of a Nature Run

A nature run is a high resolution long integration from a state-of-the-art global numerical weather prediction model. It acts as a proxy atmosphere for OSSE's from which synthetic observations from existing and future observing systems are derived. It is also used as the verification or truth data set when evaluating assimilations and forecasts which use the synthetic observations. It is usually one month or longer in length with sufficient vertical, horizontal and temporal resolution to resolve large-scale, synoptic-scale and even meso-scale weather features. It must possess realistic weather patterns and features over all time scales relevant to the OSSEs being designed.

The nature run model itself must be sufficiently different from the model used in the OSSE's to avoid producing similar climatologies. An OSSE model that has the same physics and hydrodynamics as the nature run model is referred to as an 'identical twin'. If the two models are very similar, then in regions with few simulated observations the impact of those observations would be over-estimated and in regions with many simulated observations the impact would be under-estimated. Ideally, differences between the OSSE model and the nature run model should be similar to differences between the OSSE model and the real atmosphere. Calibration experiments are used prior to undertaking any actual OSSE's to quantify and document these differences.

A nature run data set is an absolutely essential component of an OSSE. Nature runs are expensive and time-consuming to generate and validate but a single validated nature run will offer 5 to 10 years of use for a wide array of meteorological experiments performed by multiple research organizations.

List of Nature Run Datasets

+ ECMWF T213 1999 Nature Run

+ NASA Goddard FVGCM 1999 Nature Run

+ NASA Goddard FVGCM 2004 High Resolution Nature Run

+ ECMWF T511 One Year Nature Run

+ ECMWF T799 High Resolution Nature Run