RETRIEVAL OF STRATIFIED ATMOSPHERIC REFLECTIVITY AND WIND VELOCITY USING INVERSE METHODS - APPLICATION TO A VHF ST MINI-RADAR

The growing interest of measuring wind and turbulence in the lower atm osphere has led to the development of new radar systems. UHF radars, t hough being an interesting solution from a technical and logistical po int of view, have some disadvantages due to their high sensitivity to Rayleigh scattering and interference from precipitations, birds and in sects. The standard VHF ST radars were originally designed for high al titude investigations and are consequently not suited for low atmosphe re soundings. This context makes it necessary to develop the new conce pt of a VHF 'mini-radar'. But the simultaneous use of a small antenna in the VHF band, combined with a beam having a more grazing angle, res ults in an important mixing of the altitude contributions for each ran ge, a problem which is non existent with previous ST radars. Consequen tly, the atmospheric reflectivity and wind velocity profiles cannot be directly obtained and have to be treated by other methods. In this co ntext, the aim of the present work consists in the development of an a ppropriate inverse method. Two different classic methods are considere d, the least squares method and the maximum entropy method. The 'direc t problem' is first addressed, resulting in an integral description of the zeroth and first moments of the Doppler spectra. In order to perf orm various simulations to test the validity of the two proposed inver se methods in the particular case of the VHF mini-radar, a model is bu ilt for the radar which includes a set of reference atmospheric profil es. The simulations give evidence for the validity of the inversion pr ocesses. The high robustness of the least squares method always leads to significant results. But its over-determined nature results in a po or vertical resolution for the inverted profiles. Consequently this me thod is not suited to retrieving strong gradients. The maximum entropy method is intrinsically much more appropriate in terms of vertical re solution and consequently leads to valuable results, but its high sens itivity to the data noise requires some additional constraints. The pr actical efficiency of the methods is tested with real data from the mi ni-radar, and the resulting retrieved profiles are compared to those o btained simultaneously using a conventional ST radar (the 'Provence' r adar). As a result of a poor vertical resolution, the least squares me thod cannot provide a valid retrieval of the atmospheric profiles unde r real experimental conditions. Nevertheless, a preliminary inversion using the least squares method can be used as a constraint for initial izing the maximum entropy inversion process. Although this processing appears to be very efficient for the reflectivity, the retrieved profi les reveal a smoothing effect which seems to be linked to a faulty rad ar antenna radiation model. In contrast, the retrieval of wind velocit y seems to be more difficult and requires additional investigations. ( C) 1997 Elsevier Science Ltd.