Diagnosing the global hydrological cycle from routine atmospheric analyses

Abstract

AbstractThis paper discusses the extent to which atmospheric analyses produced routinely by forecasting centres can be used to diagnose the global hydrological cycle. These analyses directly represent the storage and large‐scale advection of water vapour through a combination of observations and a short model forecast. In this work, the analyses produced by the European Centre for Medium‐Range Weather Forecasts are used. Due to changes in the model formulation, which have a profound effect on the analyses, only results from three years of these initialized analyses are presented. The equations representing the conservation of water and water vapour in the atmosphere are derived and discussed, including consideration of the approximations commonly made. Diagnostics of the storage and transport of water vapour have been calculated. The results are qualitatively similar to diagnostics from other sources. In addition, with regard to attempts to derive a global relationship between the surface specific humidity and the total column water vapour, special consideration is given to the variations in the vertical distribution of water vapour. The profile of water vapour is found to be very sensitive to the competing processes responsible for the water balance in the atmosphere. Although surface fluxes of water cannot be directly derived from these analyses, a technique is introduced which puts a lower bound on the evaporation and precipitation that would be consistent with the analyses. The results of this technique are rather encouraging, and comparisons with other climatologies of surface water fluxes could provide another way of validating these atmospheric analyses. A diagnostic characterizing the time‐scale of the the small‐scale processes responsible for surface evaporation and precipitation is also introduced. This time‐scale is found to vary widely but is of the order of one day where variability in the hydrological cycle is important.