In this study the ECMWF and NCEP/NCAR reanalyses provide information on winds, temperature, and moisture. To complement the reanalyses, satellite retrievals of water vapor (NVAP), clouds (ISCCP), evaporation (HOAPS), and precipitation (MSU, GPCP) are examined. The monsoon environment is distinguished by the large role played by lateral advection in the overall moisture budget as all estimates show that regional precipitation greatly exceeds evaporation during its wet season. Remote regions where evaporation greatly exceeds precipitation export copious amounts of moisture and are found predominantly in the winter hemisphere subtropical and subsident tropical domains. Thus a large-scale hydrologic cycle is suggested that forms a critical linkage between regions greatly removed geographically.
Definitive statements can be made regarding some aspects of the mean monsoon hydrologic cycle. Most importantly, as the reanalyses agree closely in their estimates of moisture advection, it is established that a majority of latent energy in the monsoon environment is supplied by divergence in the southern Indian Ocean with a secondary yet important contribution from the Arabian Sea. Westerly transports from northern Africa are small compared to cross equatorial transports along the eastern African coast. Distributions of precipitation and evaporation confirm the patterns of moisture divergence suggested by bulk transports and net moisture convergence is positive for nearly all regions east of approximately 70E and north of the equator. The mean precipitable water distributions agree within 10% over much of the Indian and western Pacific Oceans.
Other aspects of the cycle are not represented with consistency by the various datasets. The strength of convergence varies widely in the monsoon region with the hydrologic cycle in the ERA significantly more intense than in the NCEP/NCAR reanalyses. Satellite estimates on average suggest that the NCEP/NCAR hydrologic cycle is too weak, while the ERA cycle is too strong, though variability among the satellite estimates is as large as differences between the reanalyses in many regions. Satellite estimates of evaporation from HOAPS differ substantially from the reanalyses (>50%). Given the uncertainties contained in retrievals of surface meteorology and the compounding of errors in the bulk retrieval scheme, it is not clear whether the satellite estimates are necessarily more accurate than the reanalyses.