Interannual variability of tropical ocean evaporation: a comparison of microwave satellite and assimilation results

Remote sensing methodologies for turbulent heat fluxes over oceans depend on driving bulk formulations of fluxes with measured surface winds and estimated near surface thermodynamics from microwave sensors of the Special Sensor Microwave Imager (SSM/I) heritage. We will review recent work with a number of SSM/I-based algorithms and investigate the ability of current data sets to document global, tropical ocean-averaged evaporation changes in association with El Nino and La Nina SST changes. We show that in addition to interannual signals, latent heat flux increases over the period since late 1987 range from ~ .1 to .6 mm day-1 are present; these represent trends 2 to 3 times larger than the NCEP Reanalysis. Since atmospheric storage cannot account for the difference, and since compensating evapotranspiration changes over land are highly unlikely to be this large, these evaporation estimates cannot be reconciled with ocean precipitation records such as those produced by the Global Precipitation Climatology Project, GPCP. The reasons for the disagreement include less than adequate intercalibration between SSM/I sensors providing winds and water vapor for driving the algorithms, biases due to the assumption that column integrated water vapor mirrors near surface water vapor variations, and other factors as well.

The reanalyses have their own problems with spin-up during assimilation, lack of constraining input data at the ocean surface, and amplitude of synoptic transients. A number of recent model integrations with specified SSTs are available, including “Climate of the 20th Century” integrations as well as integrations from more recent versions of the NASA GEOS climate model. These do not suffer from the “spinup” problem and evaporation from these integrations is compared to the microwave estimates and to the NCEP reanalysis.

We will also discuss the potential for improving retrievals of the near-surface specific humidity (qa) and air temperature (Ta) needed for flux calculations through the addition of moisture and temperature profile data from the SSM/T-2 and AMSU-A and –B sounders. Incorporation of information on the vertical moisture structure has enabled improved retrievals of qa in cases where the presence of moist layers aloft alters the relationship between the surface humidity and total column water vapor content. Development of an improved satellite-based qa algorithm combining SSM/I and SSM/T-2 data has enabled a reduction in the rms error in qa to 1.06 g/kg from a value of 1.12 g/kg for an algorithm incorporating only SSM/I data. Further inclusion of data from the AMSU-A sounder leads to an additional reduction in the error to 0.83 g/kg.

We will assess the current ability to quantify regional and global ocean evaporation changes, examine promising improvements afforded by additional microwave channels, and put these in the context of similar challenges to precipitation measurement and the global water cycle in general.