地理探测器方法下甘肃白龙江流域景观破碎化与驱动因子分析

Dynamics and Driving Factors of Landscape Fragmentation Based on Geo Detector in the Bailongjiang Watershed of Gansu Province

基于GIS技术、景观格局指数、主成分分析和地理探测器的方法,定量分析了1990-2014年甘肃白龙江流域的景观破碎化及其驱动因子。结果表明：1990-2014年甘肃白龙江流域景观破碎化程度先增加后降低;研究期间流域东南部的景观指数变化最为剧烈和复杂,主要是文县和武都区;人类干扰（人类活动强度和土地利用类型）和地形因子（高程、坡度和坡向）对景观破碎化空间分异的影响显著不同,其中人类干扰是景观破碎化空间分异的主要驱动因子。

The Bailongjiang watershed, located in the transitional ecotone among the Loess Plateau, Qinba Mountains and Tibet Plateau, a vital water conservation areas and ecological environment fragile areas in the upper Yangtze River with frequency geohazards, conflict between human and nature, was chosen as the study area to analyze the spatiotemporal change and driving factors of the landscape fragmentation. Based on landscape pattern index, principal component analysis, Geo Detector and platforms of GIS and RS, the spatio-temporal change and driving factors of the landscape fragmentation were quantitatively analyzed in Bailongjiang watershed of Gansu Province from 1990 to 2014. The results showed that： 1） The change regularity of landscape pattern indices showed that the degree of landscape fragmentation increased firstly and then decreased from 1990 to 2014. To be specific, Edge Density（ED） and Shannon＇s Diversity Index（SHDI） of the watershed showed the trend of increase firstly and then decrease, which indicated that the landscape of the watershed developed from irregularity to regularity in edge shape and the diversity of landscape types enhanced firstly and then weakened. Contagion Index（CONTAG） of the watershed showed the situation of decreased firstly and then increased, which suggested that reunion degree of different patch types changed from weak to strong, the higher the CONTAG, the better connectivity, otherwise, there were more small patches, representing higher fragmentation. Landscape Division Index（DIVISION） showed a slight downward trend. 2） In general, landscape fragmentation is affected by human activities, and counties or districts scale can better reflect human activities. The scale analysis of counties or districts showed that the degree of landscape fragmentation increased dramatically in Wenxian and Wudu during 1990-2002 and decreased significantly in Wenxian during2002-2014. 3） Landscape fragmentation degree of the southeastern watershed（mainly in Wenxian and Wudu）enhanced firstly and then weakened and the variation tendency was most intense and complicated. 4） Taking2014 as an example, this article explored the driving factors of landscape fragmentation based on Geo Detector.The outcome of ecological detector deemed that there were striking differences between land use types and elevation, slope, and aspect, and so did human activities intensity, nevertheless, land use types and human activities intensity had no difference, which illustrated that human disturbance（i.e. land use types and human activities intensity） and topographic factors（i.e. elevation, slope and aspect） had different effects on the spatial differentiation of landscape fragmentation. Moreover, the result of factor detector showed that human disturbance had the highest ability to explain the spatial differentiation of landscape fragmentation which meant human interference was the momentous driving factor. Finally, the result of interaction detector showed that the interaction between human activities intensity and elevation had the strongest influence on landscape fragmentation,followed by the interaction with land use types. The study aims to provide theoretical support for sustainable development in Bailongjiang watershed of Gansu Province.