地球信息科学学报 ›› 2016, Vol. 18 ›› Issue (6): 767-778.doi: 10.3724/SP.J.1047.2016.00767

• 地理空间分析综合应用 • 上一篇    下一篇

基于地理探测器的大熊猫生境适宜度评价模型及验证

廖颖1,2(), 王心源1,*(), 周俊明3   

  1. 1. 中国科学院遥感与数字地球研究所,北京 100094
    2. 中国科学院大学,北京 100049
    3. 中国天绘卫星中心,北京 102102
  • 收稿日期:2015-09-08 修回日期:2015-12-09 出版日期:2016-06-10 发布日期:2016-06-10
  • 通讯作者: 王心源 E-mail:lysky.cool@163.com;wangxy@radi.ac.cn
  • 作者简介:

    作者简介:廖颖(1990-),女,硕士生,研究方向为遥感与生态环境评价。E-mail: lysky.cool@163.com

  • 基金资助:
    国家国际科技合作专项(S2013GR0477)

Suitability Assessment and Validation of Giant Panda Habitat Based on Geographical Detector

LIAO Ying1,2(), WANG Xinyuan1,*(), ZHOU Junming3   

  1. 1. Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing 100094, China
    2. University of Chinese Academy of Sciences, Beijing 100049, China
    3. TH-Satellite Center, Beijing 102102, China
  • Received:2015-09-08 Revised:2015-12-09 Online:2016-06-10 Published:2016-06-10
  • Contact: WANG Xinyuan E-mail:lysky.cool@163.com;wangxy@radi.ac.cn

摘要:

动物生境适宜度评价对于野生动物生境保护十分重要。基于物种活动点来建模的生态位模型是目前应用最广泛的动物生境评价方法,但该方法不能直接表达生境适宜度与环境因子间具有生态学意义的数量关系。本文以雅安地区为例,提出一种新的大熊猫(Ailuropoda melanoleuca)生境适宜度评价方法,选取海拔、坡度、坡向、地形指数、距水源距离、植被类型、主食竹及距公路距离8个环境因子,引入地理探测器,在分别基于MAXENT模型和层次分析法(The Analytic Hierarchy Process,AHP)所构建生境适宜度模型的基础上,通过4个地理探测器(风险探测器、因子探测器、生态探测器和交互作用探测器)探寻大熊猫生境与各环境因子间的关系以及环境因子对大熊猫生境的影响机理,并将其预测结果与单一MAXENT模型和AHP法进行对比。结果表明:(1)AHP、AHP-Geogdetector、MAXENT和MAXENT-Geogdetector模型总体评价精度分别为85.6%、86.5%、91.3%和94.2%,kappa系数分别为0.699、0.718、0.821和0.882,AUC值分别为0.902、0.928、0.949和0.966,模型所预测的适宜和较适宜区与实际分布区重叠比分别为63.66%、61.30%、76.70%和90.10%,说明AHP-Geogdetector和MAXENT-Geogdetector模型精度均比相应的单一模型有所提高,且MAXENT-Geogdetector模型精度最高;(2)基于地理探测器的大熊猫生境适宜度评价模型能以“生境适宜度和环境因子间具有生态学意义的数量关系”的形式直接体现环境因子对动物生境利用的生态学作用,具有较好的生态学可解释性。因此,用地理探测器进行大熊猫生境适宜度评价具有较好的可行性。

关键词: 大熊猫, 地理探测器, 层次分析法, MAXENT, 生境适宜度评价

Abstract:

The suitability assessment of wildlife habitat is very important for wildlife management and protection. Niche model is the most commonly used presence-only based habitat suitability model, which cannot explicitly express the quantitative relationship between the suitability of wildlife habitat and the environmental factors, and would be insufficient to express the ecological effects of environmental factors on wildlife habitat use. In this study, a new giant panda (Ailuropoda melanoleuca) habitat assessment method based on geographical detector (Geogdetector) is proposed. A total of 8 environmental factors were selected for the suitability assessment of giant panda habitat, including the elevation, slope, aspect, topographic position index, distance from drainage system, vegetation type, staple food sources of bamboo, and distance from human settlements. Based on the initial habitat suitability index (HSI) input data calculated by Analytic Hierarchy Process (AHP) and MAXENT model respectively, we used four geographical detectors (the risk detector, factor detector, ecological detector, and interaction detector) to assess the relationship between the suitability of giant panda habitat and their environmental risk factors. Results show that the suitability assessment of giant panda habitat based on the geographical detector has relatively favorable precision and feasibility. (1) The MAXENT-Geogdetector model has higher level of performance on accuracy than the other three methods. The overall accuracy of the prediction results based on AHP, AHP-Geogdetector, MAXENT and MAXENT-Geogdetector are 85.6%、86.5%、91.3% and 94.2% respectively. The kappa coefficients are 0.699, 0.718, 0.821 and 0.882 respectively. The AUC values are 0.902, 0.928, 0.949 and 0.966 respectively. And the overlap ratios of the predicted distribution area to the actual distribution area are 63.66%, 61.30%, 76.70% and 90.10% respectively. (2) The proposed Geogdetector-based method well captures the ecological effects of environmental factors on the wildlife habitat use by indicating the quantitative relationship between the suitability of wildlife habitat and the environmental factors.

Key words: giant panda, geographical detector, analytic hierarchy process, MAXENT, habitat suitability assessment