RASS time series parametric modeling: A way to overcome the classical Fourier transform

RASS (Radio Acoustic Sounding System) is designed in order to remotely meas ure atmospheric temperature profiles by combining acoustic and radar techni ques. Usually, RASS data are processed with the same signal processing algo rithms as for wind profiling: using the classical Fourier analysis. In orde r to improve the accuracy of the temperature estimates, this paper proposes to take into account RASS time series characteristics. A parametric approa ch is studied and AutoRegressive modeling is shown to give better temperatu re estimation than the usual Fourier Transform based algorithm. (C) 2001 El sevier Science Ltd. All rights reserved.