The performances of present methodologies for data reduction are here improved, leading to a new implicit method; in addition, a new numerical concept of data reduction is presented: it includes a comprehensive 1-D finite element model which is convenient both for its capability of dealing with the most general boundary conditions and for the low computational cost as compared to the present transform-based methods.
The proposed finite element method can tackle the single- and the double-layer sensor, temperature dependent properties of the sensor and, more importantly, it can be applied when the semi-infinite assumption is no longer feasible; it proved successful in processing prototype signals for Mach numbers in the range 0.02-6. The analytical calculation of the accuracy of all the methods is also new and it prompts a guideline that is useful to design a priori an experiment on rational basis.