A CRITICAL-ASSESSMENT OF THE Q-W RELATION AND A PARAMETRIZATION RELATION FOR COMPUTING LATENT-HEAT FLUXES OVER THE INDIAN-OCEAN

Ten years (1982-1991) of upper air and ocean surface observations over the Indian Ocean from Sagar Kanya and Monex-79 data have been used to examine the relationship between the precipitable water (W) and surfa ce level humidity (Q) on an instantaneous scale. Our analysis of Q and W over the Indian Ocean shows that Q is correlated with W on an insta ntaneous scale with r=0.44, which is a statistically significant corre lation. A regression between Q and W has been fitted and it has been f ound that a fifth order polynomial yields a lowest root-mean-square (r ms) error of 1.4 g kg(-1) when compared with observed Q using an indep endent observation. The validity of earlier derived global relation be tween Q and W has been examined over the Indian Ocean. It has been fou nd that Liu's global Q-W relation gave a large rms error of 4.1 g kg(- 1) when compared with the observed instantaneous Q values over the Ind ian Ocean. The usefulness of the above derived Q-W relation and an ear lier derived relation between the monthly mean Q and W has been examin ed for the estimation of latent heat fluxes (LHF) over the Indian Ocea n using an independent observation. The LHF estimated from the bulk ae rodynamic method using all quantities available from ship observations , called the direct method (M1), has been compared with the LHF comput ed by using a derived Q-W relation (M2). The rms error between M1 vers us M2 is found to be 56 Wm(-2). The LHF estimated by Liu's Q-W relatio n, when compared with M1 gave an rms error of 155 Wm(-2), which is sug gestive of its unsuitability for the estimation of LHF over the Indian Ocean on an instantaneous basis. The difference between the sea surfa ce humidity (Q(s)) and surface level humidity (Q) has been parametrize d in terms of sea surface temperature (SST) and W, both obtainable fro m satellite sensors. This relation has then been used to compute LHF ( M3) and was compared with M1, where it was found that M1 versus M3 gav e an rms error of 58 Wm(-2). Thus, this study indicates that methods M 2 and M3 are found to be more consistent and accurate in nature, and a lso suggests that it can be further applied to the LHF estimation usin g satellite based microwave/IR measurements for W and SST on an instan taneous basis.