AMIP II Diagnostic Subproject No. 27: 
Trospospheric humidity and meridional  moisture Fluxes

Project coordinators:
Dian J. Gaffen1
Richard D. Rosen and David A. Salstein2
James S. Boyle3
Brian J. Soden4
1NOAA Air Resources Laboratory (R/E/AR)
2Atmospheric and Environmental Research, Inc., Cambridge, MA
3PCMDI, Lawrence Livermore National Laboratory, Livermore, CA
4NOAA Geophysical Fluid Dynamics Laboratory, Princeton, NJ
    Data Requirements


Tropospheric humidity and its fluxes figure prominently in both the global water and energy cycles.  For this reason, at least two diagnostic subprojects of the first AMIP considered aspects of the models' simulations of water vapor.  One of these was Diagnostic Subproject 11, "Validation of Humidity, Moisture Fluxes, and Soil Moisture in GCMs," which investigated how well the AMIP models simulated precipitable water over different portions of the globe on decadal-mean, seasonal, and interannual time scales.  Those results, and a preliminary assessment of the net meridional flux of water vapor in the AMIP model runs, are reported by Gaffen et al. (1997).

Here we propose extensions to our AMIP analyses, which were necessarily limited by the contents of the AMIP standard output archive.  The enhanced output for AMIP II, however, will allow us to consider additional moisture variables, namely, gridded specific and relative humidity and direct values of the northward flux of water vapor.  In addition, AMIP II may offer an opportunity to pursue sensitivity experiments to relate success in simulating moisture to model formulation.  The general tests performed by Gaffen et al. (1997) on the ensemble of AMIP models yielded few clear results in this regard.  Nevertheless, it is reasonable to expect that some model attributes (such as horizontal grid resolution or convective parameterization method) might influence individual AMIP II model's simulations of moisture-related fields, and we are interested in pursuing this line of inquiry; such a study would be aided if some AMIP II modeling groups undertake sensitivity experiments.

Although the principal source of observational data for AMIP II is intended to be atmospheric reanalyses, the quality of these products is problematic for water vapor (Trenberth and Guillemot 1996).  As was the case in the first AMIP, therefore, we intend to incorporate water vapor observations from a variety of sources (including radiosonde-based data sets and specialized satellite measurements) in our diagnoses of AMIP II models, and we will emphasize those regions where the data are deemed to be most reliable.


Our main objectives are the diagnosis of the climatological-mean and seasonal and interannual variability of tropospheric relative humidity and meridional moisture flux.  Two of our main results for the AMIP models were that they tend systematically to underestimate precipitable water and to overestimate decadal-mean meridional vapor flux (which we computed from the archived standard output of evaporation and precipitation).   For AMIP II we wish to investigate the following questions:

Do the biases noted by Gaffen et al. (1997) for AMIP models persist in AMIP II simulations? 
How well do AMIP II models simulate the horizontal and vertical distribution and the variability of tropospheric relative humidity, from the surface to the upper troposphere?  Does model performance vary systematically with altitude? 
How well do AMIP II models simulate the meridional flux of moisture?

a.  Relative Humidity

Using monthly mean surface and upper-air relative humidity output, we will calculate the climatological mean and measures of the seasonal and interannual variability of relative humidity from the surface to 100 hPa. These will be compared with radiosonde-based upper-air humidity observations (Ross and Elliott 1996) in the lower and mid-troposphere and with upper-tropospheric relative humidity estimates from the TIROS Operational Vertical Sounder (Soden and Bretherton 1996).  Given current interest in the ramifications of upper-tropospheric humidity variations for water vapor-greenhouse effect feedback, special attention will be devoted to that region of the atmosphere.  Comparison of the clear-sky outgoing longwave radiation with satellite observations will provide insight into how specific deficiencies in the moisture distribution influence simulations of greenhouse effect of water vapor.

Our analysis will likely involve zonal mean values and maps of particular regions in which radiosonde observations are abundant (for comparisons with radiosonde data) and global maps (for comparisons with satellite data).  While our focus will be on relative humidity, we will also investigate precipitable water and may consider temperature and specific humidity fields to determine how humidity biases are related to biases in the thermal structure of the model atmospheres.

b.  Meridional Moisture Flux

The primary moisture flux quantity we plan to diagnose is Q-flux, the zonal-mean, vertically-integrated northward transport of water vapor.  This quantity plays a major role in the global energy cycle, its latent heat accounting for approximately one-half of the total energy carried poleward by the atmosphere annually across 40 degrees N.  Beyond the climatological mean value of Q-flux, we will compute measures of the seasonal and interannual variability in Q-flux for each model during the AMIP II period.  Because the calculation of Q-flux from the archived AMIP II standard output fields may, in some cases, yield inaccurate estimates of this quantity (e.g., Williamson et al. 1996), we propose also to infer Q-flux from the archived values of evaporation and precipitation, as in Gaffen et al. (1997), to provide an independent measure of the net meridional moisture flux.

Although intermodel comparisons of Q-flux results will be important in revealing the level of agreement among the models, comparisons with observations will also be worthwhile even though reliable global analyses of moisture flux do not currently exist.  We expect to consider several observational data sets to help gauge the uncertainty in them; in particular, both reanalyses and the radiosonde-only based analyses of Oort will be used, despite each containing recognized shortcomings in moisture-related fields.  Our main focus will be on Q-flux and its temporal variability, but we also plan to examine the fields of v and q used to construct Q-flux to identify the sources of major discrepancies in Q-flux, either among models or between models and observations.

Data Requirements

We will require the following model output fields, organized as in Appendix A of AMIP Newsletter No. 8.

Table 1 (Upper-air low frequency)
Northward wind speed
Air temperature
Specific humidity
Relative humidity
Pressure surface below ground
Mean product of northward wind and specific humidity

Table 2 (Single-level low frequency)
Surface air temperature
Surface pressure
Total precipitation rate
Precipitable water
Surface specific humidity
TOA clear-sky longwave radiation (method II)
Surface evaporation plus sublimation rate.

Table 5 (Fixed geographic fields)
Model topography


Gaffen, D.J., R.D. Rosen, D.A. Salstein, J.S. Boyle, 1997: Evaluation of  tropospheric water vapor simulations from the Atmospheric Model Intercomparison Project. J. Climate, 10, 1648-1661.

Ross, R.J., and W.P. Elliott, 1996: Tropospheric water vapor trends over North America: 1973-93. J. Climate, 9, 3561-3674.
Soden, B.J., and F.P. Bretherton, 1996: Interpretation of TOVS water vapor radiances  in terms of layer-average relative humidity: Method and climatology for the upper, middle and lower troposphere. J. Geophys. Res., 101, 9333-9343.

Trenberth, K.E., and C.J. Guillemot, 1996: Evaluation of the atmospheric moisture and hydrological cycle in the NCEP reanalyses, NCAR Tech. Note (NCAR/TN-430+STR), 308 pp.

Willliamson, D.L., D.H. Bromwich, and R.-Y. Tzeng, 1996: Further discussion on simulation of the modern Arctic climate by the NCAR CCM1. J. Climate, 9, 1669-1672.

For further information, contact the AMIP Project Office (

LLNL Disclaimers