当前位置:网站首页>地理探测器原理介绍
地理探测器原理介绍
2022-07-02 21:32:00 【BetterQ.】
空间分异性是地理现象的基本特点之一,地理探测器是探测地理要素空间分异性,以及揭示其驱动因子的一种方法,由王劲峰等人提出。
地理探测器是探测和利用空间分异性的工具,其主要包括4部分:分异及因子探测、交互作用探测、风险区探测、生态探测。根据地理探测器的原理及应用领域,本文使用第一种探测器来分析地形地貌对河网发育的影响及二者间的关联性。
(1)分异及因子探测器: 探测研究对象Y的空间分异性,以及探测各个因子X在多大程度上解释了Y的空间分异性。其解释程度用q值来度量,表达式为:
式中,h=1,2,⋯,L为变量Y或因子X的分层,即分类或分区;N_h和N分别为层h和全区的单元数;σ_h2和σ2分别是层h和全区的Y值的方差;SSW和SST分别为层内方差之和与全区总方差。q的取值范围为[0,1],q值越大说明Y的空间分异性越明显;如果分层是由变量X生成的, q值越大则说明自变量X对属性Y的解释力越强,反之则越弱。在极端情况下,当q值为1时,表明变量X能够完全控制Y的空间分布;当q值为0时,则表明变量X与Y没有任何关系,q值表示变量X解释了100×q%的Y。
q值的一个变换满足非中心F分布:

式中,λ为非中心参数,Y ̅_h为层h的均值。根据式2.20可以用来检验q值是否显著。
(2)交互作用探测: 识别不同风险因子Xs之间的交互作用,即评估因子X1和X2共同作用时是否会增加或减弱对因变量 Y 的解释力,或这些因子对Y的影响是相互独立的。评估的方法是首先分别计算两种因子 X1 和 X2 对 Y 的q 值 : q(X1) 和 q(X2),并且计算它们交互(叠加变量 X1 和 X2 两个图层相切所形成的新的多边形分布,图2)时的q值: q(X1 ∩ X2) ,并对q(X1)、q(X2)与q(X1 ∩ X2)进行比较。
(3)风险区探测: 用于判断两个子区域间的属性均值是否有显著的差别,用t统计量来检验:
(4)生态探测: 用于比较两因子X1和X2对属性Y的空间分布的影响是否有显著的差异,以F统计量来衡量:
具体内容见文献:王劲峰,徐成东.地理探测器:原理与展望[J].地理学报,2017,72(01):116-134.
配套的还有用于地理探测器的软件包,是在excel中使用,可以点击链接直接下载使用,里面附有示例数据:地理探测器软件包
边栏推荐
- Research Report on right-hand front door industry - market status analysis and development prospect forecast
- Import a large amount of data to redis in shell mode
- treevalue——Master Nested Data Like Tensor
- ~90z axis translation
- Redis分布式锁故障,我忍不住想爆粗...
- Market trend report, technical innovation and market forecast of China's Micro pliers
- Etcd Raft 协议
- I drew a Gu ailing with characters!
- Construction and maintenance of business websites [4]
- Analysis of enterprise financial statements [3]
猜你喜欢

MySQL learning record (4)

MySQL learning notes (Advanced)
![[hands on deep learning]02 softmax regression](/img/47/eb67ec2c51f6bb7d6b2879b36e769d.jpg)
[hands on deep learning]02 softmax regression

Structure array, pointer and function and application cases

Report on investment development and strategic recommendations of China's vibration isolator market, 2022-2027
![[shutter] shutter layout component (opacity component | clipprect component | padding component)](/img/6b/4304be6a4c5427dcfc927babacb4d7.jpg)
[shutter] shutter layout component (opacity component | clipprect component | padding component)

System (hierarchical) clustering method and SPSS implementation

Read a doctor, the kind that studies cows! Dr. enrollment of livestock technology group of Leuven University, milk quality monitoring

How to prevent your jar from being decompiled?
![[question brushing diary] classic questions of dynamic planning](/img/31/fcd8230f809d6178f11e7095c1ef94.jpg)
[question brushing diary] classic questions of dynamic planning
随机推荐
Market trend report, technical dynamic innovation and market forecast of China's low gloss instrument
The neo4j skill tree was officially released to help you easily master the neo4j map database
Report on investment development and strategic recommendations of China's vibration isolator market, 2022-2027
ctf-HCTF-Final-Misc200
Makefile: usage of control functions (error, warning, info)
~90z axis translation
[shutter] shutter layout component (wrap component | expanded component)
Capacity expansion mechanism of ArrayList
AES encryption CBC mode pkcs7padding filling Base64 encoding key 32byte iv16byte
Research Report on market supply and demand and strategy of China's plastic pump industry
7. Build native development environment
[shutter] statefulwidget component (pageview component)
[shutter] shutter layout component (Introduction to layout component | row component | column component | sizedbox component | clipoval component)
China microporous membrane filtration market trend report, technological innovation and market forecast
Research Report on right-hand front door industry - market status analysis and development prospect forecast
MySQL learning record (2)
Research Report on plastic antioxidant industry - market status analysis and development prospect forecast
How is LinkedList added?
Basic knowledge of tree and binary tree (detailed illustration)
Common authority query instructions in Oracle