Novel method for inversion of microphysical properties of clouds using Raman lidar data

Aerosol–cloud–precipitation interactions are important in the balance of Earth’s radiation budget. To further explore the relationship between clouds and precipitation, and to improve operational weather modification, it is necessary to study the microphysical parameters of liquid water clouds. Here, an inversion method that uses a back propagation (BP) neural network based on a genetic algorithm (GA), namely a GABP, is proposed to invert cloud microphysical parameters using ground-based dual-field-of-view (FOV) Raman lidar data. To verify the feasibility of the method, long-term continuous observations were conducted in the Liupan Mountains (China). Results revealed that the proposed inversion method using the GABP is feasible for retrieving the liquid water content (LWC) and the cloud droplet effective radius after training a large number of data measured simultaneously by the Raman lidar and a microwave radiometer. When inverting LWC, the root mean square error (RMSE) of the GABP algorithm was found in the range 0–0.005, whereas the RMSE of the BP algorithm fluctuated in the range 0–0.01. It was evident that the GABP algorithm yields better inversion results and finer detail. When maintaining other variables and comparing the inversion results of signals in the inner and outer FOVs, the RMSE of the inner FOV signal was within 0.005 at near-ground heights (i.e., <2 km), whereas the outer FOV signal exceeded 0.005 at certain heights. This study developed a feasible solution for detecting characteristic cloud microphysical parameters using a Raman lidar, which could be used to study aerosol–cloud–precipitation interactions, and thereby have considerable practical importance for improving artificial rainfall operations.