Hi, Zhongqi

Usually I made grid for interpolation using R software which is open source programming language(http://www.r-project.org/).The grid is kind of point data, each grid can be generate using sample method. After that I import the point to sGeMS as a irregular shaped grid. 

Hi Bingwei:

Thanks very much for your reply. It's very helpful. The R software is totally new for me. Can you be more specific? Like how to import the grid from R to SGeMS, is SGeMS need to be complied? And as far as I know, the only two data format can be imported into SGeMS are binary and ASCII. I'm just very confused about the process.

SGeMS has a binary file type but if you want to edit it yourself I would advise doing it in the ASCII format which works as follows (it's GEOEAS format):

* Assuming I have a point file with X,Y,Z and Variable columns the header will have to be:

name_of_the_set

4

X

Y

Z

Variable

11     5     0      0.35

11     4.9  1      0.29

10.3 5      2     0.31

...       ...    ...    ....

1) So the first row is a string with the name of the set (any name but avoid uncommon characters and perhaps white space). 2) The second row is the number of variables. In this case "X,Y,Z,Variable" makes 4. 3) Than you have a row for each variable with only a string (the name of the variable). 4) Finally you have the data organized by columns (the first column is X, the second is Y, and so on...).

For GRID files (the previous example is for points). You'll have:

name_of_the_grid

1

Variable

1

2

3

4

5

6

7

8

9

The same logic as points but this time you don't have to give X,Y,Z coordinates. The location is given by the number of blocks. This file could be a 3x3x1 grid organized in the following way:

7   8   9

4   5   6

1   2   3

So the first value is in X=0,Y=0; the second value is in X=1,Y=0; third in X=2,Y=0; and so on in lines of X, increasing Y followed by Z.

So you can either put the points in SGeMS or put a masked grid where you put null data values in a regular grid to mask the nodes where no values exist.

Hi pedro:

Thanks very much for your patiently answer my question. But what I'm trying to do is to build irregular grid, which is not like a cubic region with certain number of cells in X,Y,Z directions.  I need the grid to show the structure of formation underground, like the picture I posted above, so the cubic flat grid is not enough. 

I've never used a base version of SGeMS that allows stratigraphic grids. I'm guessing that if exists is a plugin. That said I couldn't even say if it's a free plugin. You should take a look at this: http://www.ar2tech.com/products

They have a proprietary version of SGeMS which has what you need. I believe that at least one of the founders of the company is also one of the developers for SGeMS so I'm guessing his academical research is entwined with his company career. However I'm not sure of this.

That said I was suggesting ways of "emulating" an irregular grid. For actual stratigraphic grids I do not know how to do it. On the side please consider that the variography study should be done in the same space as the one on which you intend to estimate or simulate. Most people will "flatten" the stratigraphic grid in order to achieve the horizontal layers from which to calculate the variogram. Notice that I doubt SGeMS has a variogram engine capable of dealing with any mesh that is not based on 8 vertices by nodes (like triangular, or other polygon irregular meshes) where all nodes are adjacent.

I assume you have a irregular shape file, for example 2D boundary of your study area named studyArea.shp.

shp

Hi Bingwei:

Thanks so much for your help. I tried your way, everything is OK until I use the mask grid as interpolation grid, the results are not like yours. I attached the screenshot to show my results. There should be something wrong with it.