The work is concerned with the inversion of lidar data into the aerosol particle size distribution (APSD). The main difficulty of this problem is the discrete character of lidar observations in terms of the width of observation bands, and a limited number of observation wavelengths. The method of mean ordinates was developed and numerically tested for inverting optical data of this kind. The method makes it possible to retrieve, under certain conditions, APSD from lidar data at two and even one wavelength. When doing this, the retrieval accuracy is comparable with that of direct APSD measurements. The method also allows the determination of the aerosol component presenting some problems for direct observations, namely, the large-particle part of APSD (r ³ 5 mm).

The method of mean ordinates overcomes certain shortcomings peculiar to the so-called look up tables, such as a limited set of initial models and the solution criterion. In contrast, the method of mean ordinates is based on a wide variation of aerosol parameters, which results in a more accurate selection of acceptable solutions. The most probable solution is defined as an acceptable solution that is found to be the closest to the mean over the whole ensemble of acceptable solutions.

The method of mean ordinates was tested in many numerical experiments and by using the experimental data of the Shoreline Environment Aerosol Studies (SEAS) 2000. When inverting SEAS lidar data, two sets of initial models were constructed. The first set was based on aerosol data available in literature. The second one used direct SEAS observations over the aerosol (simultaneously with lidar, at the lidar site). It should be noted that the lidar data are related to a distance of 300 m offshore from the lidar. This correlates with the average between the experimental APSD at 5 and 15 m above the ground at the experimental site. The figure presents the comparison of the inversion results with the direct SEAS data for APSD. It is seen that the inversion results for the second set show a noticeable improvement as to those for the first set.

Figure. Inversion results by the method of mean ordinates for 23h06m, April 25.

1, inversion with the use of the first set of initial parameters; 2, measured APSD at the 5 m height; 3, measured APSD at the 15 m height; 4, inversion with the use of the second set of initial parameters; 5, curves of minimal and maximum ordinates of the ensemble of acceptable solutions for the first set oinitial parameters; 6,7 curves of minimal and maximum ordinates of the ensemble of acceptable solutions for the second set of initial parameters.

This was the reason why the second set was used for the retrieval of the APSD profile along the lidar beam. The analysis of the results revealed the position of ocean sprays along the beam trajectory and the change of the aerosol structure before the sprays, in sprays, and beyond them.

Supplementary URL: