As noted earlier, presently available global climatologies leave much to be desired in terms of the spatial and temporal resolution of net water flux over the ocean. While separate climatologies for precipitation [ Legates and Wilmott, 1990] and evaporation [ Ebensen and Kushnir, 1981; Hsiung, 1986] have been developed, there have been no attempts to compute a global E-P climatology since Baumgartner and Reichel [1975]. In part this is due to the distrust of the process of differencing two numbers with sizeable uncertainties. However, failure to attempt such a climatology leaves even the most fundamental features of the system unspecified. There is a great need for definition of at least the general amplitude and patterns of surface water fluxes as input for ocean models and for comparison with evolving observational techniques.
Precipitation estimates over the sea have been developed by compiling statistics on the frequency, type and intensity of rainfall from the ``present weather'' observations of mariners. Tucker [1961] originated this approach, in which coast and island rainfall data are used to develop empirical relations with the local marine observations. These regressions are then applied over the open ocean. Reed and Elliot [1979] prepared precipitation maps for the North Atlantic based on rain frequency with an additional seasonal correction. Dorman and Bourke [1979, 1981] use rainfall frequency, type and intensity as well as an additional correction for temperature, in order to achieve adequate dynamic range in the tropics. As millions of ship observations are available over the open ocean, this technique eliminates the need to extrapolate across the ocean basins from coast and island data alone. Thus, much more detail can be seen in their charts compared to those of Baumgartner and Reichel.
There is also hope that the variability in oceanic precipitation can be determined by use of the present weather codes. D. Cayan [ (personal communication, 1994] reports that many 5 squares in the North Atlantic and Pacific have more than 200 observations per month. While systematic trends in the data and changes in reporting procedures raise some difficulties, he finds significant correlation of marine precipitation patterns with nearly land data.
Evaporation is computed from the standard ``bulk formula'' relating air--sea exchange of moisture to boundary-layer humidity gradients and the 10 meter wind speed. A number of basin scale [ Bunker, 1976; Isemer and Hasse, 1987] and global scale [ Hsiung, 1986; Ebensen and Kushnir, 1981] climatologies have been produced. Calculations have also been made of variability in the latent heat flux [ Cayan, 1992].