S. Godin-Beekmann

Long term stratospheric DIAL (Differential Absorption Lidar) ozone lidar (Light Detection and Ranging) measurements have been performed at the Observatoire de Haute-Provence (OHP) since 1985 and as part of the Network for the Detection of Stratospheric Changes (NDSC) since 1991. This paper provides a detailed description of the DIAL lidar instrument implemented at OHP and a discussion of the ozone retrieval analysis. The instrument includes the detection of atmospheric nitrogen Raman scattering wavelengths, which is required for ozone measurements in the presence of strong volcanic aerosol loading. A comprehensive evaluation of the error budget in the 10–50 km altitude range is performed with particular emphasis on the bias associated with background and volcanic stratospheric aerosol. This bias is evaluated using ancillary measurements of the aerosol size distribution obtained from 1991 to 1999 after the Mount Pinatubo volcanic eruption. Results show that the bias associated with background aerosols is smaller than 1% above 13 km while the bias due to volcanic aerosols reaches a maximum of 5% above 15 km in the DIAL Raman ozone retrieval. In background aerosol conditions the total accuracy of the DIAL ozone vertical distribution at OHP ranges from 5% below 20 km to 15–30% above 45 km with a vertical resolution varying from 0.5 km to 5 km.

An intercomparison of ozone differential absorption lidar algorithms was performed in 1996 within the framework of the Network for the Detection of Stratospheric Changes (NDSC) lidar working group. The objective of this research was mainly to test the differentiating techniques used by the various lidar teams involved in the NDSC for the calculation of the ozone number density from the lidar signals. The exercise consisted of processing synthetic lidar signals computed from simple Rayleigh scattering and three initial ozone profiles. Two of these profiles contained perturbations in the low and the high stratosphere to test the vertical resolution of the various algorithms. For the unperturbed profiles the results of the simulations show the correct behavior of the lidar processing methods in the low and the middle stratosphere with biases of less than 1% with respect to the initial profile to as high as 30 km in most cases. In the upper stratosphere, significant biases reaching 10% at 45 km for most of the algorithms are obtained. This bias is due to the decrease in the signal-to-noise ratio with altitude, which makes it necessary to increase the number of points of the derivative low-pass filter used for data processing. As a consequence the response of the various retrieval algorithms to perturbations in the ozone profile is much better in the lower stratosphere than in the higher range. These results show the necessity of limiting the vertical smoothing in the ozone lidar retrieval algorithm and questions the ability of current lidar systems to detect long-term ozone trends above 40 km. Otherwise the simulations show in general a correct estimation of the ozone profile random error and, as shown by the tests involving the perturbed ozone profiles, some inconsistency in the estimation of the vertical resolution among the lidar teams involved in this experiment.