A Simple Method of Deriving Three-Dimensional Temperature Fields Using Remotely Sensed and In Situ Data for Application to Numerical Hydrodynamic Models of Estuaries and Bays

Abstract

This paper describes the subjective interpolation method (SIM) for generating three-dimensional temperature distributions from remotely sensed sea surface temperature (SST) fields. SIM incorporates MATLAB-based cloud removal software and a method of generating synthetic temperature profiles based on observations. This approach depends on the human facility for recognizing patterns in complex images. Three-dimensional temperature fields produced by SIM are compared to analogous fields based on optimal interpolation (OI) methods by using temperature fields interpolated by the two methods to initialize a baroclinic coastal ocean circulation model. The initial SST surface fields from both methods have a bias of less than ?0.5°C and rms errors of less than 1.5°C. After running for 48 h, the bias and rms errors for the OI simulations are 0.3° and 1.2°C, respectively, whereas the same errors for the SIM run are 0.7° and 0.9°C. The OI and SIM approaches can be combined to allow preprocessing of SST data in three steps: 1) bad pixels can be marked by a human user (quality control), 2) automated OI methods can be used to produce valid SST fields based on available reference data, and 3) the SST field can be projected into the water column using a combination of observed and synthetic profiles.