3D temperature modeling for the South China Sea using remote sensing data

Abstract

The South China Sea's (SCS) seasonal, large-scale temperature cycle is governed to a large extend by the monsoon. This phenomenon modulates the large-scale circulation, transport and mixing as well as the exchange processes with the Pacific Ocean and the East China Sea. Also, significant variations in net surface heat flux will contribute to the large-scale, seasonal temperature cycle. As a result, a seasonal mixed layer temperature cycle of over 6oC occurs in the northern SCS regions and between 2oC and 4oC in the southern regions. Over the central SCS temperature stratification is observed throughout the year, while over the shallow northern and southern regions atmospheric forcing and large-scale transport will attribute to a seasonal breakdown of the stratified system. The objective of this study is to assess the large-scale three-dimensional temperature cycle of the SCS and to develop a corresponding hydrodynamic model that is resolving the monsoonal response. Due to the significant spatial and temporal scales, sea level anomalies observed by satellite altimetry and Sea Surface Temperature (SST) observed by satellite radiometer play an essential role in this study, both to assess the SCS physical system and for modelling applications. The model is setup using the Delft3D-FLOW hydrodynamic modelling package and applies an orthogonal spherical-curvilinear and boundary fitted grid in the horizontal. In the vertical a sigma-layer approach is applied. In the deep SCS regions the model depth is truncated based on a reduced depth approach. For surface heating the so-called Ocean heat flux model of Delft3D-FLOW is used. At the open model boundaries water level and lateral transport forcing is applied. The model does not resolve tidal forcing. An extensive sensitivity analysis is performed, with model forcing and validation data both for a climatological year and for the year 2000. The models temperature accuracy is subsequently improved by assimilating remotely sensed SST data using a nudging method. On seasonal scales, the model represents the large-scale transport, surface heating and stratification with reasonable accuracy. Without SST nudging a mean difference of 1.75oC is observed with respect to validation data. By nudging SST the mean difference decreases with 15% to 1.5oC.