Role of winds in estimation of ocean heat storage anomaly using satellite data

Abstract

The heat storage anomaly of the upper ocean can be estimated using altimeter data, based on the thermal dynamics equation. In this study, we analyzed both nonthermal and thermal steric height measured by altimetry, and removed nonthermal steric height from altimeter measurements to calculate the heat storage anomaly from thermal steric height alone in the global oceans, but with a focus on the Pacific Ocean. After examining nonthermal steric height (salinity, waves, and wind), we believe that wind stress is a major component to the change in sea surface height. Blended TOPEX/Poseidon and ERS‐1/2 altimeter 1° × 1° sea surface height anomaly from January 1993 to October 2000 were used to calculation of the heat storage anomaly. In addition, we used expendable bathythermograph (XBT) data to the global oceans to determine a reference heat storage, from which the altimeter estimation of the heat storage was compared. The wind stress curls, from scatterometer data obtained from January 1993 to October 2000, were used for analyzing the wind effects on sea surface height variation. The correlation coefficients between heat storage from altimeter data (HAlt′) and that from the XBT (HXBT′) are larger than 0.6 in the global oceans, except in the regions of 20°N in the eastern Pacific, the eastern tropical Pacific, the tropical Atlantic, and the western tropical Indian Ocean. The root mean square (RMS) of the differences (HAlt′ − HXBT′) has large values in the Gulf Stream and the Kuroshio Extension regions as well as the eastern and western tropical Pacific, and eastern Indian Ocean. The wind‐induced sea surface height variation exhibits strong annual and interannual cycles, so the empirical orthogonal function (EOF) method was employed to extract the cycles. The annual cycle has two modes, whereas the interannual cycle is characterized by El Niño events. Wind stress can affect the estimation of the heat storage through mechanisms derived from quasi‐geostrophic and barotropic theory. We calculated the correlation coefficient between the differences (HAlt′ − HXBT′) and wind stress curl. The result shows that El Niño Southern Oscillation (ENSO) related wind stress curl anomalies are responsible for estimating discrepancy in the regions west of South and North America and tropical oceans. The high correlation suggests that in these regions the barotropic and linear theory is the dynamic mechanism underlying the wind effects on heat storage estimation. Furthermore, we focus on baroclinic signal in sea level resulting from Ekman upwelling in the tropical oceans. The effect of wind‐forcing has the most important role in sea level change during the 1997–1998 El Niño event. For instance, the heat storage anomaly (Q′) due to the strong coastal upwelling off the west coast of Peru closely relates to sea surface temperature (SST), wind‐forcing, remnant sea level height anomaly (Δη′) after removing the contributions of the heat, salt, and waves to the sea level height, and Ekman depth (DEK). However, the Q′ in the west coast of America and western Pacific warm pool areas did not closely relate to SST, wind‐forcing, Δη′, and DEK.