Methodology for the independent calibration of Raman backscatter water-vapor lidar systems.

We present a method for the independent calibration of Raman backscatter water-vapor lidar systems. Particular attention is given to the resolution of instrumental changes in the short and the long terms. The method reposes on the decomposition of the instrument function, which allows the lidar calibration coefficient to be re-expressed as the product of two terms, one describing the instrumental transmission and detection efficiency and the other describing the wavelength-dependent convolution of the Raman backscatter cross sections with the instrument function. The origins of changes in instrument response necessitate the experimental determination of the system detection efficiency. Two external light sources for calibration are assessed: zenith observation of diffuse sunlight and a xenon arc lamp. The results favor use of the diffuse-sunlight measurement but highlight the need for simultaneous sunphotometer measurements to constrain modeled aerosol optical properties. Quantum mechanical models of the Raman cross sections are described, and errors in determining the cross sections and their convolution with the instrument function are discussed in detail. The calibration coefficients deduced by using the independent method are compared with coefficients deduced from Vaisala H-Humicap radiosonde measurements. These results agree to within current calibration errors (15%, unconstrained aerosol parameters), and a change in calibration coefficient following instrument modification is reproduced satisfactorily. Results from modeling and intercomparison studies are extended to estimate the calibration accuracy and the precision of the diffuse-sunlight method with constrained modeled aerosol parameters. Changes in the calibration coefficient in the short and the long terms should be resolved to 4(6)% and 6(9)%, respectively, which is comparable or better than the precision of existing dependent methods of calibration. The reduction of the absolute calibration error remains an outstanding issue for all calibration methods.