Measuring Heat Exchange Processes at the Air--Water Interface from Thermographic Image Sequence Analysis

In this thesis a novel technique for estimating heat transfer at the free air water interface is presented. For the first time spatially resolved heat flux and transfer velocity measurements are available with a high temporal resolution. The statistical properties of the transfer processes are deduced and the parameters characterizing them established. Based on this analysis a second way to estimate the heat flux is presented. These techniques are based on thermal image sequences on which a motion analysis is performed. The motion is modelled in a general parameterization and physically motivated intensity changes can be incorporated by means of linear partial differential equations. In the presented framework the parameters of physical processes described by such differential equations can be estimated in multidimensional data. These general equations of motion are solved simultaneously by a method of least squares. To do so algorithms are developed that allow for unbiased estimates taking the structure of the noise into account. Methods from robust statistics are employed to correctly solve the estimation problem regardless if the data is corrupted by outliers. The relevance of the developed techniques to other scientific applications is shown. In an accuracy analysis confidence bounds of the proposed algorithms are established and limitations revealed. Following an examination under controlled laboratory conditions in the Heidelberg Aeolotron, the techniques are successfully applied at an international field campaign.