Regionalization of the matching degree of water, soil, and heat resources in Central Asia based on ecosystem services using PSO-SOFM neural network

Regionalization of the matching degree of water, soil, and heat resources is of great significance for regional agricultural planning. The long-term unreasonable management of water, soil, and heat resources has caused regional resource shortages and environmental problems in Central Asia, which seriously threatens agricultural production in this region. However, few studies have investigated the regionalization patterns of the matching degree of water, soil, and heat resources in Central Asia. In this study, the spatio-temporal patterns of four ecosystem services, including vegetation carbon sequestration, soil conservation, water supply and conservation, and biodiversity conservation, were quantified by using remote sensing data. Combined with the Particle Swarm Optimization (PSO) and Self-Organizing Feature Map (SOFM) neural network, the regionalization of the matching degree of water, soil, and heat resources was examined. The relationships among various eco-environmental factors of different matching degree zones were assessed using Spearman's rank correlation analysis. The effects of temperature and precipitation on ecosystem services in Central Asia were analyzed by using partial correlation analysis. The results showed that the ecosystem services were generally high in the southeast while low in the northwest, decreasing from the mountains to the oases and the deserts. The four ecosystem services showed different degrees of change from 2000 to 2015 in Central Asia. Areas with significantly reduced vegetation carbon sequestration and soil conservation accounted for 84.81% and 84.82% of Central Asia, respectively, and areas with significantly reduced water supply and conservation and biodiversity conservation accounted for 69.48% and 19.8% of Central Asia, respectively. However, the ecosystem services from water supply and conservation and biodiversity conservation increased in some areas. The PSO-SOFM neural network model performed well in the regionalization of the matching degree of water, soil, and heat resources in Central Asia. The matching degree of water, soil, and heat resources in Central Asia can be divided into five categories with 21 sub-categories according to the patterns of ecosystem services. At the spatial scale, there were significant differences in the ecosystem services among different matching degree zones. Precipitation was the most important limiting factor affecting the ecosystem service values and matching degree, whereas the effects of temperature and soil properties were less important. At the temporal scale, the areas with a significant positive correlation between precipitation and ecosystem services were larger. The significant effect of temperature on ecosystem service values was mainly concentrated in ecological sensitive zone of northern Kazakh steppe and semi-desert, ecological fragile zone of desert in Central Asia, ecological sensitive zone of central semi-desert in Central Asia and ecological sensitive zone of semi-desert in Badghyz and Karabil. In other regions, temperature and precipitation were not the main factors affecting ecosystem services. Changes in the ecosystem service values may be related to land use types. Combined with the ecological and geographical conditions of different matching degree zones, this study provides useful information for the development and utilization of water and land resources, agriculture and animal husbandry development, and environmental protection in Central Asia.