Abstract
The variability in the meridional Ekman heat flux estimated using wind data from four different sources is examined. The wind vectors are obtained from the European Remote Sensing (ERS), Quick Scatterometer (Quikscat), and Special Sensor Microwave Imager (SSM/I) satellite instruments and from the National Centers for Environmental Prediction (NCEP) model. The datasets range over a period of 10 years except for the Quikscat, which spans the period between 1999 and 2003. The comparison of the annual mean of the zonally integrated Ekman heat flux shows some discrepancies. In comparing the four sources, the differences increase from the tropical regions toward the equator. The annual mean of the meridional Ekman heat flux is consistently smaller when estimated with the ERS data. The correlation analysis shows that ERS and the other sources have a better agreement in the tropical regions, with correlations between 0.6 and 0.8, while in the extratropical regions the correlation is 0.4. The SSM/I, NCEP, and Quikscat winds lead to better correlations, between 0.7 and 1 in the extratropical regions. The western side of the north Indian Ocean is a site where all sources are very well correlated to each other. The variability in the Ekman heat flux is determined by changes in the temperature and wind stress fields. A combination of digital filters was used to quantify the role of several regions in the frequency–zonal wavenumber spectrum of the wind in establishing the observed Ekman heat flux. The Ekman flux component that is obtained from the product of the long-term mean wind and the temperature dominates in the low latitudes of the Atlantic Ocean. Its fractional covariance reaches 0.6 in the Atlantic, in the Pacific Ocean it is at most 0.3, and it is negligible in the Indian Ocean. The temporal variability of this heat flux component is only due to the temperature variability, because the mean winds were used. Other Ekman heat flux components are obtained from the product of the filtered wind anomalies and the temperature. These components include several bands of propagating signals (Rossby waves) and have fractional covariances that are larger in the Pacific and Indian Oceans, while in the Atlantic they can explain at most 20% of the total variance. All wind sources show a shift in the variability regime around 15° of latitude, with the mean and large-scale prevailing over meso- and small-scale variability within the Tropics and vice versa in the extratropical regions.
Corresponding author address: Olga T. Sato, Instituto Oceanográfico da Universidade de São Paulo (IOUSP), Praça do Oceanográfico, 191, São Paulo, SP 05508-120, Brazil. Email: olga@io.usp.br
