Evaluation of Impacts from Real Observations and Their Simulated Counterparts Using the Historical Observing System Simulation Experiment Methodology

The accurate characterization of the meteorological and oceanographic environment at tactical scales (5 km or less horizontal scales from 0 to 12 hours of forecast lead time) is an essential need for the U.S. Navy. The domains the Navy operates in are often constrained by limited data, which make it more difficult to form accurate initial conditions for the initialization of numerical weather prediction (NWP) models. Traditionally, observing system simulation experiments (OSSEs) have been performed to evaluate new sensors and platforms designed to help fill data gaps. In an OSSE, a high resolution nature run is used to simulate all observations (including simulated data from the existing observing system as well as data from new sensors and platforms), which are then assimilated by the target data assimilation system. The effectiveness of the new observations are expressed as measures of how close analysis and forecast output are to the nature run. Unfortunately, traditional OSSEs are prohibitively expensive to run for the Navy due to the need of a high resolution nature run, simulation of all observations, and calibration of the observation and background errors.

An alternative to the traditional OSSE methodology has been proposed in which new observations to be evaluated are drawn from an alternate, independent NWP system which is run over a specific time period of the historical record. Unlike the traditional OSSE methodology, the existing global observing system does not need to be simulated in this historical OSSE technique; instead, existing observations come from the historical record. Leveraging the historical record also reduces the required effort in validating the reference state (the counterpart to the nature run in the traditional OSSE), which serves as the information source used to draw the new observations as well as validation state for the experiment. This research investigates the use of the historical OSSE technique by comparing estimated impacts of forecast improvement from simulated observations from new, non-traditional observing platforms to their real-world counterparts using the Coupled Ocean-Atmosphere Mesoscale Prediction System (COAMPS®) as the target NWP system and the Weather Research and Forecast (WRF) model as the reference state.