数据挖掘算法在高寒草地退化驱动因素相关性分析中的应用

高寒草地的退化受到众多自然、人为活动的影响,退化与驱动因素之间的耦合关系复杂。本文以青海省称多县为研究区,提取2005—2014年归一化植被指数(NDVI,normalized difference vegetation index)时间序列数据集,结合温度、降水、社会经济因素,运用基于数据挖掘的提升度算法进行相关性分析,研究影响高寒草地退化指标与表示高寒草地退化指标之间两者的关系。本文采用提升度算法针对3个等级的NDVI、可食量、植株高度与相应的温度、降水、鼠害和放牧强度之间的关系进行分析,能够更精确地分析各驱动因子在不同等级取值范围下对草地不同等级退化的贡献率,发现驱动因素与草地退化之间的影响关系并不是单向的,而是到达一定的程度时会出现逆向影响关系。本研究得到如下结论:(1)草地植被覆盖度低与气温和降水存在负相关;(2)草地可食量低与气温和人口存在负相关,与牲畜存在正相关;(3)地上植株高度低与牲畜呈现正相关。

Application of data mining algorithm in correlation analysis of the driving factors for alpine grassland degradation

The degradation of alpine grassland is affected by a range of natural and human activities, and the coupling relationship between degradation and its driving factors is complex. To study the relationship between the influencing factors and indicators for alpine grassland degradation, this paper refers to Duo County as the research area, extracts the NDVI time series data from 2005 to 2014, combines the temperature, precipitation and socio-economic factors, and uses the data mining-based lift caluation in the correlation analysis. The relationships between the three levels of NDVI, food availability or plant height and the corresponding temperature, precipitation, rodent or grazing intensity were examined, so as to more accurately analyze the contribution rate of each driving factor to different grades of grassland degradation. The conclusions are: (1)The low vegetation coverage in grassland is negatively correlated with temperature and precipitation. (2)The low food availability in grassland is negatively correlated with temperature and population, and positively correlated with livestock. (3)The low plant height is positively correlated with livestock.