Key points are not available for this paper at this time.
Abstract Probability distribution functions of tropospheric water vapour in the tropics are shown to be commonly bimodal. This bimodality implies sharp gradients between dry and moist regimes in space and time. A method of testing for and quantifying bimodality is introduced. Using this method, the bimodality of water vapour is surveyed in satellite and in situ observations, as well as in global model re‐analysis data and simulations. The bimodality suggests that the radiative drying time after an injection of moisture by convection is short (1–2 days) compared to a homogenizing time, whether physical (mixing) or mathematical (averaging). It is shown that the local bimodality found in cloud‐model simulations and in situ point measurements disappears with modest time averaging (18 h and 200 km), but then reappears on the global‐scale, where dry and moist regions are separated so widely that synoptic‐ and large‐scale mixing times exceed the drying time‐scale. Large discrepancies exist in the ability to reproduce the global‐scale bimodality by global model re‐analysis and simulations. Copyright © 2003 Royal Meteorological Society
Zhang et al. (Wed,) studied this question.