Aerosol Layer Discrimination using Laser Radar and Genetic Algorithms

Abstract

A technique has been developed to retrieve the height of the top of the aerosol layer from Micro Pulse Lidar (MPL) datasets. The technique combines first derivative estimates of normalized relative backscatter profiles with genetic algorithm refinements. The genetic algorithm is used to explore the gradient profiles to produce temporally coherent results. I. INTRODUCTION The distribution of aerosols over altitude and time has important effects on air quality, pollution,radiative forcing and climate, and rainfall patterns. Studies of aerosols also provide important information on atmospheric dynamics and transport. Thus atmospheric studies often require information on aerosol quantities and distributions. One important parameter for these studies is aerosol layer height. Lidar measurements of atmospheric backscatter may be used to track the height of the aerosol-rich layer over time. In this study, aerosol distribution data were obtained using the Connecticut State University (CSU) Lidar Collaboratory's Micro Pulse Lidar (MPL) system. The MPL is used to provide aerosol profiles for a variety of applications including air quality assessment and pollution control, climate modeling and studies of local atmospheric dynamics (1), (2), (3), (4). The CSU Lidar Collaboratory MPL is a Type 4 System from Sigma Space Corporation which monitors elastic backscatter at 527 nm. The system is eyesafe and thus may be operated autonomously. The MPL is a useful tool for boundary layer studies as it is capable of providing data in both daytime and nighttime conditions. Laser light pulses at 2500 Hz are transmitted vertically out of a beam-expansion telescope and the resulting backscatter is detected by a photon counting avalalanche photodiode. Detected intensity provides informa- tion on aerosol optical properties while timing of the scattered pulse return provides the altitude of the scatterer. Data used for this study were measured with an altitude resolution of 15 meters and a time interval of one minute. The data were range-corrected and also corrected for instrument artifacts and calibrations. The resulting datasets consist of time, altitude and normalized relative backscatter (NRB) signal intensity. These data are represented as images in which altitude is plotted on the vertical axis and time on the horizontal axis. The pixel value at each image pixel represents the NRB signal intensity. Examples of NRB datasets collected at New