Thank you BrickFielder for indicating this paper. Unfortunately it doesn't help to solve my problem of estimating the water content from *surface* data. It's only by exception that also data at higher levels are available, like e.g. the above mentioned soundings from Manaus which I used for checking the estimation procedure. As you indicated, estimating the water content from surface data requires that some assumptions are made, like e.g. the scale height of 1.5 km. Any ideas on how this scale height could be estimated (instead of assuming a fixed value) with the available surface data (temperature, dew point temperature, pressure and relative humidity)? Is there e.g. a relationship/approximation between the height of the LCL (which can be estimated from surface and dew point temperatures) and the scale height?