Institute of Meteorology and Climate Research

MIPAS HDO/δD observations help to detect and diagnose biases of processes controlling tropospheric humidity in atmospheric general circulation models (GCMs)


The figures show the global distributions of δD, i.e. the depletion of atmospheric water vapor with respect to deuterium relative to the Standard Mean Ocean Water reference, in the upper troposphere. MIPAS observations (left) have been compared to global fields from LMDZ (right). In the top row, annual means of δD at 250 hPa are compared, while in the bottom row, seasonal variations (June/July/August - December/January/February) at 250 hPa are compared. The model underestimates both the meridional gradient and the seasonality of the δD distribution.

A comparison between LMDZ and a large number of satellite, ground-based and in-situ data sets on the ratio HDO/H2O covering the troposphere from the surface to 250 hPa has been performed, in order to gain insight into model biases in processes controlling subtropical tropospheric humidity. The comparison between LMDZ and all observational data were consistent in the following findings: in all comparisons, the amplitude of seasonal variations in the isotopic composition at all levels in the subtropics and in midlatitudes was underestimated; all comparisons showed that the observed meridional isotopic gradient was underestimated, and all comparisons to satellite data showed that the contrast between dry and convective tropical regions was underestimated. Comparison of LMDZ to six other GCMs demonstrated that all analysed GCMs have a similar problem. In a second paper it was demonstrated that the moist bias found in many GCMs in the mid and upper troposphere most frequently results from an excessive diffusion during vertical water vapor transport. This conclusion was only possible by comparing both relative humidity and isotopic fractionation of water vapor from observations and model data.

For further details see