[ArcGIS小白记]Kriging插值

ArcToolBox>>Spatial Analyst Tools>>Interpolation>>Kriging

引用来源:http://bbs.esrichina-bj.cn/ESRI/viewthread.php?tid=55815

地统计与确定性插值的最大区别在于,地统计插值引入了概率模型,即地统计插值认为从一个统计模型不可能完全精确地得出预测值,所以在进行预测时,应该给出预测值的误差,即预测值在一定概率内合理。

通常所说的地统计插值是指克里格插值法(Kriging)
Z
(s)=μ(s)+ε(s)
S表示不同的位置点,可以是用经纬度表示的空间坐标。
Z(s)是该位置点的属性值。μ(s)为确定趋势值,ε(s)
为自相关随机误差。

当要考虑多个协同变量的情况下,可采用协克里格插值法(Co-Kriging).则其计算公式将变为:
Zj(s)=μj(s)+εj(s)
表示的是第j个变量的情况。在协克里格中,只对主变量进行预估,但将在插值预估时引入不同变量间的随机误差项εj(s)的交叉相关性值,从而构建协同克里格模型。

克里格插值是一个最优的无偏估计法。

获得预测图并不要求数据呈正态分布。但当数据呈正态分布时,克里格插值法将是无偏估计法中效果最好的一种方法。

因此,在进行克里格插值前,可先对非正态分布的数据进行转换,包括Log对数转换,Box-Cox转换,使之呈正态分布,然后再进行插值。
根据样点数据统计特征的不同可将克里格分成多种不同的插值法:

当样点数据是二进制值时,用指示克里格插值法进行概率预测;

对样点数据进行了未知函数变换后,可用该变换函数进行析取克里格插值;

当样点数据的趋势值μ(s)是一个未知常量时,用普通克里格;

当样点数据的趋势可用一个多项式进行拟合,但回归系数未知时,用泛克里格插值法;

当样点数据的趋势已知时,用简单克里格插值法;

其中最常用的是普通克里格与泛克里格插值法;当加入了协变量进行插值时,则叫作协同普通克里格插值法和协同泛克里格插值法。

同反距离权重插值法IDW一样,克里格插值法同样可以表示为:
Z(x0)=
∑λi Z(xi)
Z(x0)
为待估点的值, Z(xi) 为待估点周围的已知样点值,λi为第i个已知点的权重

所不同之处在于,IDW的权重为待估点与已知样点间距离的u次幂的倒数,而克里格的权重值不仅考虑待估点与已知样点、已知样点之间的距离,还考虑了其空间分布的方位。通过半变异函数来赋权重值。

24,597 thoughts on “[ArcGIS小白记]Kriging插值

  1. I used to beI was recommendedsuggested this blogwebsiteweb site throughviaby way ofby means ofby my cousin. I amI’m now notnotno longer surepositivecertain whetherwhether or not this postsubmitpublishput up is written throughviaby way ofby means ofby him as no onenobody else realizerecognizeunderstandrecogniseknow such specificparticularcertainpreciseuniquedistinctexactspecialspecifiedtargeteddetaileddesignateddistinctive approximatelyabout my problemdifficultytrouble. You areYou’re amazingwonderfulincredible! Thank youThanks!

  2. YesterdayThe other dayToday, while I was at work, my sistercousin stole my iPadiphoneapple ipad and tested to see if it can survive a thirtyforty40twenty five2530 foot drop, just so she can be a youtube sensation. My iPadapple ipad is now brokendestroyed and she has 83 views. I know this is completelyentirelytotally off topic but I had to share it with someone!

  3. GreetingsHiyaHey thereHeyGood dayHowdyHi thereHello thereHiHello! I know this is kinda off topic however ,neverthelesshoweverbut I’d figured I’d ask. Would you be interested in exchangingtrading links or maybe guest writingauthoring a blog articlepost or vice-versa? My websitesiteblog goes overdiscussesaddressescovers a lot of the same subjectstopics as yours and I feelbelievethink we could greatly benefit from each other. If you happen to beyou might beyou areyou’re interested feel free to sendshoot me an e-mailemail. I look forward to hearing from you! AwesomeTerrificSuperbWonderfulFantasticExcellentGreat blog by the way!

  4. Right nowCurrentlyAt this time it seemssoundslooksappears like BlogEngineMovable TypeDrupalExpression EngineWordpress is the besttoppreferred blogging platform out thereavailable right now. (from what I’ve read) Is that what you’reyou are using on your blog?

  5. NiceExcellentGreat blog here! Also your websitesiteweb site loads up fastvery fast! What hostweb host are you using? Can I get your affiliate link to your host? I wish my websitesiteweb site loaded up as fastquickly as yours lol

  6. Thank you very much for your writing and your article and I wish you continued success. Please try to produce such valuable content in the following processes. In addition to this article, I am very happy to see your unique content.

  7. Thank you very much for your writing and your article and I wish you continued success. Please try to produce such valuable content in the following processes. In addition to this article, I am very happy to see your unique content.

  8. Hatay evden eve nakliyat için kurumsal çözümler sunan firmamız siz değerli müşterilerine her daim güvenli taşınmalar sağlamıştır. Hatay evden eve nakliyat firmamız yıllar önce Hatay merkezde kurularak burada faaliyetlerine başlamıştır. Bunun yanı sıra ülkemizin her bölgesine de taşımacılık yapabilmektedir. Hatay evden eve nakliye ihtiyaçlarınızda her daim sizlerin yanındayız.

Leave a Reply

Your email address will not be published. Required fields are marked *