Research On Water Poverty In Gansu Section Of Silk Road Economic Belt Based On Random Forest And Geodetector

The water resources crisis will not only trigger a series of shocks such as food shortages,immigration conflicts,and financial turmoil,but also bring huge risks to human life and livelihood security.The traditional water resources evaluation is confined to the water resources system,and the relationship between water resources,ecology,society,politics,economy are not well discussed.Water poverty theory is based on a comprehensive multi-disciplinary basis.Starting from general poverty theory,and it links people's ability to use water resources embodied in economic development and social security with the development,utilization,and management of water resources.The Water Poverty Index can help water resource managers to quantitatively assess the current status of regional water resources and monitor changes in the amount of available water resources and the changes in socioeconomic factors related to water resource acquisition and use,which also can provide the necessary information on establishing a water security system.The cities in the Gansu section of the Silk Road Economic Belt in the arid and semi-arid areas of the west are in an important strategic opportunity period of "opening to the west",and they are also one of the most prominent regions of poor water problems of China.Analyzing the spatial and temporal distribution and impact factors of water poverty in Gansu section can provide theoretical basis for urban construction and water resources management.Based on the theory of water poverty,this paper establishes an evaluation system and grading standard based on urban water poverty in the arid regions of Northwest China.The WPI method and random forest classification method are used to evaluate urban water poverty,and the spatial and temporal characteristics of water poverty distribution in Gansu section are analyzed.The results and applicability of the two methods are summarized.The geodetector method was introduced into impact factor analysis to detect impact factors that have a greater driving force for water poverty differentiation.This study draws the following conclusions:(1)The water poverty situation in Gansu has been significantly improved.The WPI value of the Gansu section obtained by using the weighted sum method increased from 0.305 in 2000 to 0.354 in 2017.The number of extreme water-poor cities decreased,and the number of moderate and mild water-poor cities increased significantly;From 2012 to 2017,the number of cities with "level 1" water poverty(extremely poor)decreased to 0,while the number of cities in the "level 4"(good)and "level 5"(excellent)status increased significantly.(2)The results of the WPI method and the random forest method have strong consistency,and both methods can be applied to the evaluation of water poverty at the regional scale.The former are a weighted comprehensive evaluation based on entropy weight,which can ensure the result is larger and richer in information.The latter can handle high-dimensional data,with less human intervention and faster training speed and it is suitable for processing data sets with a large number of unknown features.(3)The spatial pattern of water poverty in Gansu section shows a change from "relatively mixed distribution of water poverty at various levels" to "gradually decreasing from southeast to northwest".(4)Jiayuguan,Jinchang and Lanzhou belong to “capable longboard” type.Jiuquan and Zhangye are “resource longboard” type.Tianshui,Pingliang and Qingyang are weak “resource longboard”.Wuwei,Baiyin and Dingxi are "low quality balanced" type.(5)The enhancement of social and economic strength plays an important role in improving the situation of water poverty.The primary factor affecting the spatial differentiation of water poverty in Gansu has changed from the amount of water used for food to the proportion of R&D expenditure.(6)There are large regional differences in the main factors that affect the changes in the evaluation values of water resource poverty in various cities.Jiuquan is greatly affected by the intensity of chemical fertilizer application.Jiayuguan corresponds to the water production modulus,Zhangye corresponds to the groundwater resource per capita,Jinchang and Lanzhou correspond to the water consumption of 10,000 GDP,Baiyin corresponds to the water saving irrigation rate,Dingxi,Tianshui,Pingliang The proportion of the city and Qingyang corresponding to other water sources.