Retrieval of aerosol particle size distribution using an improved lévy flight and circle chaos sparrow search algorithm

The aerosol size distribution is a crucial metric for evaluating aerosol optical characteristics and determines the direct and indirect radiative forcings of aerosols. Traditional methods for inversion of aerosol size distribution often suffer from ambiguities and limitations. To address these shortcomings, this paper introduces a method for inferring aerosol volume size distribution utilizing an improved Lévy flight and chaotic sparrow search algorithm (ILCSSA). The algorithm incorporates Circle chaotic mapping onto the basic Sparrow Search Algorithm (SSA) to obtain a high-quality initial population and employs a Lévy flight strategy to enhance population diversity. Additionally, the process of updating the population’s positions is optimized to improve algorithm accuracy. To validate the feasibility of the proposed method, the measured aerosol optical depth (AOD) data obtained from a Precision Solar Radiometer (PSR) sun photometer in Shouxian are utilized and a series of comparisons were conducted among the Sparrow Search Algorithm, Standard Particle Swarm Algorithm, Improved Particle Swarm Algorithm, and Spider Wasp Optimization Algorithm. The results demonstrate a significant performance advantage for the ILCSSA, evidenced by an average reduction of 50 <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="m1"><mml:mi>%</mml:mi></mml:math> in Sum of Squared Errors (SSE) and 36 <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="m2"><mml:mi>%</mml:mi></mml:math> in Root Mean Squared Error (RMSE) when compared to the other four algorithms. Additionally, the AOD obtained by ILCSSA had a correlation coefficient of 0.9748 with the original AOD data. Furthermore, we analyzed the aerosol volume size distribution in Shouxian under conditions of good air quality, moderate pollution, and mild pollution. The proposed method holds significant reference value in the field of aerosol volume spectrum inversion.