Comparison of Aura MLS water vapor measurements with GFS and NAM analyses in the upper troposphere/lower stratosphere

Aura MLS v1.5 water vapor measurements taken over a roughly two year period are compared with GFS and NAM analyses valid over the same period. The Aura data are interpolated to every 50 mb and compared to the model analyses at levels between 300 and 50 mb. Because the AURA data represent a swath 7 km wide by 200 km long, averaging is performed over a similar sized region in the analysis data, which have a grid spacing of 1 degree latitude and longitude for the GFS and 12 km for the NAM. Comparisons are made for each three-month season.

It is found that a moist bias is present in both model analyses compared to the satellite estimates, and this bias is worst at the highest levels. Probability density functions of the data suggest a much narrower distribution in the Aura data. In particular, it appears both model analyses contain points that have much higher water vapor mixing ratios than found by satellite. Averages taken over subsets of the GFS domain including the tropics, midlatitudes and polar regions all reveal similar biases.