Test dataset

A stochastic model with a von Karman auto-covariance function with nu = 0.5, variance = 1 and correlation lengths c_x =100m and c_y=10m is provided with any05.mat. The model has a size of 20m*20m and a sampling interval of 0.1m. (201points*201points)

A subsampled dataset of this stochastic model in a n*m matrix is presented in input.mat. This dataset is sampled with a sampling interval of 0.4m.

To proceed kriging interpolation, load this files in matlab and set the path for the 'vebyk_functions' folder (see the readme file inside the folder). Then use the 'inputmatrix' function to convert the input dataset to the format desired by 'vebyk':

[inputformat] = inputmatrix(input,0.4,0.4);

Now process kriging interpolation with:

[output,errorvariance] = vebyk(inputformat,[0.1 0.1],16,10,0,0.98,100,0,1)

after a while, the calculation will finish and the results are stored in 'output', a 3*40401 matrix. Every line represents one point. The first row contains the x-coordinate, the second the y-coordinate and the third is the estimated value at this point. The result can be displayed with:

matrixdisplay(output,201,201);

Compare to the stochastic model with

figure
imagesc(any05)




% Rolf Sidler 10.10.2003 rolf.sidler@gmx.ch