Adaptive Estimation of ATMS Observation Uncertainty to Improve Atmospheric Prediction

We are currently in an exciting era with an unprecedented number of novel observations for data assimilation systems. This abundance of new observations necessitates finding optimal ways to use them. One key challenge is the estimation of their uncertainty. For satellite observations, most operational centers assume they are well calibrated and that their observation errors do not vary spatially or temporally. In this study, we relax both of those assumptions for radiance observations and demonstrate on the ATMS (Advanced Technology Microwave Sounder) instrument. The ATMS, with 22 channels, is capable of providing temperature and humidity profiles in all-weather conditions, as well as precipitation information.

We estimate the observation uncertainty using a proven Desroziers method, which uses the residuals of observation-minus-analysis (O-A) and observation-minus-background (O-B). We adaptively estimate both the uncertainty (in terms of standard deviation) of each channel, as well as the error correlation of each channel, for ATMS observations across various geographic regions and time. Finally, the impact of adaptively estimating the ATMS observation uncertainties, both homogeneously and heterogeneously represented in time and/or space, is evaluated in the US NAVy Global Environmental Model (NAVGEM) forecasts initialized from analyses that made use of these estimates.