The major heat components for air-sea interface include shortwave, longwave, latent, and sensible heat fluxes. Among them, shortwave and latent heat fluxes are the dominate factors over tropical oceans. In these regions, latent heat fluxes are comparable to the net shortwave fluxes. Currently, satellite remote sensing techniques start to provide global estimates of latent heat fluxes. Validations using in situ measurements are critical for the satellite estimates.

In this study, latent heat fluxes have been retrieved using brightness temperature measurements by the TRMM satellite for 1998 and 1999. The available ground measurements are from the NOAA Environmental Technology Laboratory (ETL) research ships. They are from cruise measurements for the projects such as the Joint Air-Sea Monsoon Interaction Experiment (JASMINE), the Nauru 99 (NAURU99), the Kwajalein Experiment (KWAJEX), the Pan-AMerican Climate Study (PACSF99), and the buoy service in the North Pacific (MOORINGS). The five cruises provided three latent heat estimates: direct covariance (COV), inertial-dissipation (ID) methods, and the bulk algorithm (BULK) using ship measured meteorological variables.

The TRMM Microwave Imager (TMI) pixels were collocated with each of the ship measurements to be within 10 minutes and 20 km. Each ship measurement has 1 to 20 matched TMI pixels. These matched pixels were run through the COARE algorithm of Fairall et al (1996), using satellite estimated meteorological parameters such as surface temperature, humidity, and wind speed, to calculate the latent heat flux. The mean values of these parameters were compared to the ship measurements. The bias (rms) errors of these meteorological parameters are 0.31 (1.69) m/s for wind speed, -0.36 (1.28)K for sea surface temperature, -0.04 (1.35) K for surface air temperature, and 0.50 (1.49)g/kg for air specific humilities.

The satellite calculated latent heat fluxes have bias (rms) of -9.74 (37.08) W/m**2 when compared to the ship BULK results. They are 1.53 (54.30) W/m**2 and 0.27 (50.0) W/m**2 when compared to the ship COV and ID results. This is very promising because the bias (rms) are 8.21 (38.9) W/m**2 and 10.0 (43.6) W/m**2 between BULK and COV, and between BULK and ID. The correlation coefficients between satellite estimated fluxes and surface BULK, COV, and ID are 0.56, 0.30, and.0.43, respectively. The main error sources in satellite flux retrievals are the uncertainties in air humidity and wind speed. The ship motion, flow distortion, and the contaminating effects of marine environment in the ship measurements produce major uncertainties in the in-situ flux estimates. Despite all these obstacles, the satellite measurements provide reasonable accuracy with global coverage.