摘要
In this paper, the self-adaptive artificial fish swarm algorithm (SAAFSA) is used to optimize the coarse graining of segment numbers, which are used in the Lempel-Ziv complexity algorithm. This approach improves the Lempel-Ziv complexity (LZC) algorithm of equal probability coarse graining. As a case study, the complexities of monthly series of groundwater depth were analyzed at seven farms in the Hongxinglong Administration. GIS technology was used to create spatial distributions of monthly groundwater depths. A projection pursuit model based on the SAAFSA was established and used for complexity attribution analysis at selected farms with different degrees of complexity. The three selected farms, Hongqiling, 852, and Youyi, each represent a certain degree of complexity. Further analysis shows that precipitation, evaporation, temperature, and human activities are the primary factors that cause complexity variations in local groundwater depth. The results reveal the evolution of the complexity characteristics of local groundwater depth and provide scientific evidence for the need to effectively allocate regional water resources. Additionally, the proposed method can be applied in complexity analyses of other hydrologic features, as well as in research regarding nonlinear time series in economic, engineering, medical, and signal analyses.