In Situ Cloud Sensing with Multiple Scattering Lidar: Simulations and Demonstration

Abstract

Due to the spatially inhomogeneous nature of clouds there are large uncertainties in validating remote sensing retrievals of cloud properties with traditional in situ cloud probes, which have sampling volumes measured in liters. This paper introduces a new technique called in situ cloud lidar, which can measure extinction in liquid clouds with sampling volumes of millions of cubic meters. In this technique a laser sends out pulses of light horizontally from an aircraft inside an optically thick cloud, and wide-field-of-view detectors viewing upward and downward measure the time series of the number of photons returned. Diffusion theory calculations indicate that the expected in situ lidar time series depends on the extinction and has a functional form of a power law times an exponential, with the exponential scale depending on the distance to the cloud boundary. Simulations of 532-nm wavelength in situ lidar time series are made with a Monte Carlo radiative transfer model in stochastically generated inhomogeneous stratocumulus clouds. Retrieval simulations are performed using a neural network trained on three parameters fit to the time series of each detector to predict 1) the extinction at four volume-averaging scales, 2) the cloud geometric thickness, and 3) the optical depth at four averaging scales. Even with an assumed 20% lidar calibration error the rms extinction and optical depth retrieval accuracy is only 12%. Simulations with a dual wavelength lidar (532 and 1550 nm) give accurate retrievals of liquid water content and effective radius. The results of a mountain-top demonstration of the in situ lidar technique show the expected power-law time series behavior.