An Evaluation of the Atmospheric Water Budget Components over the Southwest Pacific Using Multiple Datasets, with Special Emphasis on Evaporation

An evaluation of the atmospheric water budget components over the Southwest Pacific using multiple datasets, with special emphasis on evaporation

Authors: Ning Li and Gilles Bellon

An analysis is made of the atmospheric water budget components (i.e., moisture flux divergence Dv, evaporation E and precipitation P) over the oceans of the Southwest Pacific (95°E~225°E and 15°N~70°S) for the period 1979-2013 using multiple datasets. Specifically, the datasets include the blended OAFlux E (Eo), reanalysis ERA-Interim E (Ei) and Dv, satellite-based HOAPS E (Eh), and observation-based GPCP P. Focus has been placed on E due to its nature of being a derived variable and hence highly error prone. To appreciate latitudinal differences, the region is split into three zones, that is, 15°N-15°S (Z1), 15°S-45°S (Z2) and 45°S-60°S (Z3). The analysis shows that considerable uncertainties exist in the interannual fluctuations and magnitudes (but not direction) of the trend estimate of E. Moreover, assuming the tendency term is negligible, Dv as computed from reanalysis specific humidity and wind fields does not balance the difference between the three datasets of E and P. The main finding regarding E can be summarized as follows: 1) E over Z1 exhibits two moderate peaks at solstices but E over Z2 and Z3 have peaks (troughs) in austral winter (summer), an apparently counter-intuitive result that can be explained by positive seasonal deviation in zonal winds and an enhanced air-sea specific humidity gradient to the west of Australia, giving rise to vigorous E during austral winter, 2) the strongest interannual variability and upward trend is seen in Eh presumably because the Mt Pinatubo eruption in 1991 has caused anomalously low Eh for at least 3 years, 3) dropping the years of Eh affected by the eruption and for the common period 1988-2008, all three E datasets show statistically significant albeit very different estimates of upward trend in Z1 and Z2, up to an average of 0.3%±0.07%/yr and 0.4±0.08%/yr, respectively, 4) the spatial configurations of the E datasets are reasonably similar; E over the tropical Pacific Ocean has the lowest inter-dataset similarity, 5) the best E estimate for the region as a whole is 3.4 mm/day. In summary, the three E datasets have reasonable agreement on seasonal cycle but poor agreement on interannual variability and trends. The regional trend is larger than the global oceanic E trend. It appears that accurately characterizing evaporation is still a pressing challenge faced by the research community.