Variational Aircraft Temperature Bias Correction in the NCEP's GSI Analysis System

Various studies have noted that aircraft temperature data are biased. In this study, variational aircraft temperature bias correction is incorporated in the Gridpoint Statistical Interpolation (GSI) data assimilation system at National Centers for Environmental Prediction (NCEP). Several bias models, which are aircraft tail number dependent, are investigated. The results show that the aircraft temperature bias correction cools down the atmosphere analysis around 200mb, and improves the analysis and forecast fits to the radiosonde data. Overall, the quadratic vertical velocity bias model performs better than other bias models tested here.

Two other issues are also discussed in this study. One is the lack of tail numbers for AIREP aircraft data, and the impact of AIREP data is presented. The other is the missing nearest second information for the observation time used in the aircraft vertical velocity calculation. In addition to the finite-difference method, a tensioned-splines method is used to obtain more continuously smooth aircraft vertical velocity and mitigate the missing time information.