A Bayesian estimate of CT2 from sonic anemometer data

We present a method of infering the temperature structure parameter (CT2) from sonic anemometer data using a Bayesian analysis. Specifically, we first create a time series of temperature increments from the original sonic temperature time series and use the expected form of the temperature structure function to deduce the correlations between the increments in the series. By assuming that the increments are distributed according to a multivariate Gaussian distribution, the conditional probability for the increment time series for a given CT2 can be deduced. We then used Bayes' rule to find the conditional probabilty for CT2 given the observed increment time series and take the most probable value of CT2 as our measurement from the sonic anemometer data. Finally, we show a comparison of these CT2 values with those simultaneously obtained from a laser scintillometer.