https://wenku.baidu.com/view/546e107bf01dc281e53af0e7.html
https://wenku.baidu.com/view/cfd2d9cca1c7aa00b52acbb4.html
γ ( d ) = { 0 , d = 0 C 0 + C ( 3 2 ∗ d a − 1 2 ∗ d 2 a 2 ) , 0 < d ≤ a C 0 + C , d > a \gamma(d) = \begin{cases} 0,\quad d=0\\ C_{0}+C(\frac{3}{2}*\frac{d}{a}-\frac{1}{2}*\frac{d^2}{a^2}), \quad 0<d \leq a\\ C_{0}+C, \quad d>a\\ \end{cases} γ(d)=⎩⎪⎨⎪⎧0,d=0C0+C(23∗ad−21∗a2d2),0<d≤aC0+C,d>a
γ ( d ) = { 0 , d = 0 C 0 + C ( 1 − e d a ) , d > 0 \gamma(d) = \begin{cases} 0,\quad d=0\\ C_{0} +C(1-e^{\frac{d}{a}}), \quad d>0 \end{cases} γ(d)={ 0,d=0C0+C(1−ead),d>0
γ ( d ) = { 0 , d = 0 C 0 + C ( 1 − e d 2 a 2 ) , d > 0 \gamma(d) = \begin{cases} 0,\quad d=0 C_{0} +C(1-e^{\frac{d^2}{a^2}}), \quad d>0 \end{cases} γ(d)={ 0,d=0C0+C(1−ea2d2),d>0
γ ( d ) = { d a , 0 < a ≤ 2 \gamma(d) = \begin{cases} d^a,\quad 0<a \leq 2\\ \end{cases} γ(d)={ da,0<a≤2
γ ( d ) = { l o g d \gamma(d) = \begin{cases} log_{d} \end{cases} γ(d)={ logd
γ ( d ) = { 0 , h = 0 C 0 ( > 0 ) , h > 0 \gamma(d) = \begin{cases} 0,\quad h=0\\ C_{0} (>0) , \quad h>0 \end{cases} γ(d)={ 0,h=0C0(>0),h>0
Z ( s 0 ) ^ = ∑ i = 1 n λ i Z ( s i ) \hat{Z(s_0)} = \sum_{i=1}^n \lambda_{i}Z(s_i) Z(s0)^=i=1∑nλiZ(si)
普通克里金插值的条件:对于整个空间来说,空间中的每一个变量都有着相同的期望E和方差σ2。
E [ z ( x , y ) ] = c V a r [ z ( x , y ) ] = σ 2 E[ z(x,y) ] = c \\ Var[z(x,y)] = \sigma^2 E[z(x,y)]=cVar[z(x,y)]=σ2
空间内容一点都等于常量c 和随机偏差量R(X,Y)之和
z ( x , y ) = E [ z ( x , y ) + R ( x , y ) ] = c + R ( x , y ) z(x,y) = E[z(x,y) + R(x,y) ] = c+R(x,y) z(x,y)=E[z(x,y)+R(x,y)]=c+R(x,y)
R ( x , y ) = V a r [ z ( x , y ) ] = σ 2 R(x,y) = Var[z(x,y)] =\sigma^2 R(x,y)=Var[z(x,y)]=σ2
https://en.wikipedia.org/wiki/Variogram
γ ( i j ) = 1 2 E [ ( Z i − Z j ) 2 ] \gamma(ij)=\frac{1}{2}E[(Z_i-Z_j)^2] γ(ij)=21E[(Zi−Zj)2]
Zi 、Zj 分别为空间中位置i、位置j 对应的变量值。
由于普通克里金插值的特性 :空间中任一点都等于恒定空间期望C 与随机偏差R 之和
Z i − Z j = R i − R j Z_i - Z_j = R_i - R_j Zi−Zj=Ri−Rj
公式变形
γ ( i j ) = 1 2 E [ ( R i − R j ) 2 ] = 1 2 E [ R i 2 ] + 1 2 E [ R j 2 ] − E [ R i ∗ R j ] = σ 2 − c o v ( R i , R j ) \gamma(ij)=\\\frac{1}{2}E[(R_i-R_j)^2]\\ =\\\frac{1}{2} E[R_i^2]+ \frac{1}{2} E[R_j^2]-E[R_i*R_j] =\\ \sigma^2 - cov(R_i,R_j) γ(ij)=21E[(Ri−Rj)2]=21E[Ri2]+21E[Rj2]−E[Ri∗Rj]=σ2−cov(Ri,Rj)
相似空间距离表述
d i j = d ( Z i , Z j ) = d [ ( x i , y i ) , ( x j , y j ) ] = ( x i − x j ) 2 + ( y i − y j ) 2 d_ij = d(Z_i,Z_j) = d[(x_i,y_i),(x_j,y_j)] = \\ \sqrt{(x_i-x_j)^2+(y_i-y_j)^2} dij=d(Zi,Zj)=d[(xi,yi),(xj,yj)]=(xi−xj)2+(yi−yj)2
本文代码及可视化代码均放在 gitee 码云上 欢迎star & fork