Synthetic radiance simulation and evaluation for a Joint Observing System Simulation Experiment

Abstract

Since 2006, a Joint Observing System Simulation Experiment (OSSE) project has been established at the United States Joint Center for Satellite Data Assimilation. The work provided here documents the simulation of synthetic satellite observations from the OSSE Nature Run (NR) and the evaluation of the simulated results. The Community Radiative Transfer Model was used to produce synthetic satellite observations, which will be assimilated in the Gridpoint Statistical Interpolation system. Synthetic radiances were evaluated through a comparison with real observations and model simulations obtained from National Center for Environmental Prediction (NCEP) Global Forecast System (GFS) 6 h forecast fields. For both IR and microwave sensors, we determined that the bias and the standard deviation of the synthetic radiances were in good agreement with real observations. At the NR initial time, the simulated Advanced Microwave Sounding Unit‐A (AMSU‐A) radiances reproduced observed AMSU‐A interchannel correlations and symmetric angular‐dependent features. However, the asymmetric angular‐dependent bias, mainly related to AMSU‐A instrument polarization misalignment, cloud not be simulated properly. The simulated GOES‐12 Sounder radiances indicated that the error characteristics of atmospheric temperature sounding channels were similar to those from the NCEP operational GFS analysis system, and those biases of moisture and surface channels were approximately 2 K. To mimic real observational errors, random (Gaussian distribution) errors with the mean and variance estimated by the differences between observations and GFS fields were added to the synthetic radiances, resulting in a 20–30% increase in the standard deviations of the synthetic radiances.