----Original Message----
> (Surfer) + KRIGING, produces excellent,
believable contours the great
>majority of the time(VM, get kriging if at all possible)
>VM is by far better software, if it had Kriging it would be hands
>down the winner, however, you can't always get good contours with VM, I'd
>say about 10% of the time.
>
Colin:
I like to remind other readers that VM has many other functions than interpolating grids and creating contour lines, however, your concern about choosing an appropriate interpolation technique is well founded.
We are currently working on a long overdue white paper that covers many of the end-user queries about choosing appropriate interpolation techniques. The many 3d images really help to identify the issues. It includes examples using the new Kriging module slated for the next VM release (release date not yet available). The new kriging system includes Simple, Ordinary and Universal Kriging, error estimation grids, full variogram design tools and nested models. It will also have a 'beginner' and 'advanced' mode so that anyone can get useful results. I will make a posting to the MapInfo-L when this is available on our web site.
Just to give a hint of the issues that will be discussed, let me point out that although kriging most often gives results that look intuitively correct it cannot model local high or low points. Suppose several water wells are drilled but none intersect the top (apex) of the acquifer dome. If kriging (without nugget effect) is used to grid this data it will not be able to model the apex which rises above the highest values in the sample data set. TIN and Natural Neighbour techniques which use slope functions can model this dome. Our tests using NN or TIN techniques on evenly spaced linear contours show a perfectly sloping surface without steps. However, contorted contour lines will generate areas where flat spots or steps are visible due to the geometry of the underlying TIN (triangular irregular network) used to determine slopes.
Hope this will help!
James Harvie
Vertical Mapper Product Manager
Northwood Geosience Ltd.
43 Auriga Drive, K2E 7Y8
Nepean, Ont, Canada
613-224-2020
>majority of the time(VM, get kriging if at all possible)
>VM is by far better software, if it had Kriging it would be hands
>down the winner, however, you can't always get good contours with VM, I'd
>say about 10% of the time.
>
Colin:
I like to remind other readers that VM has many other functions than interpolating grids and creating contour lines, however, your concern about choosing an appropriate interpolation technique is well founded.
We are currently working on a long overdue white paper that covers many of the end-user queries about choosing appropriate interpolation techniques. The many 3d images really help to identify the issues. It includes examples using the new Kriging module slated for the next VM release (release date not yet available). The new kriging system includes Simple, Ordinary and Universal Kriging, error estimation grids, full variogram design tools and nested models. It will also have a 'beginner' and 'advanced' mode so that anyone can get useful results. I will make a posting to the MapInfo-L when this is available on our web site.
Just to give a hint of the issues that will be discussed, let me point out that although kriging most often gives results that look intuitively correct it cannot model local high or low points. Suppose several water wells are drilled but none intersect the top (apex) of the acquifer dome. If kriging (without nugget effect) is used to grid this data it will not be able to model the apex which rises above the highest values in the sample data set. TIN and Natural Neighbour techniques which use slope functions can model this dome. Our tests using NN or TIN techniques on evenly spaced linear contours show a perfectly sloping surface without steps. However, contorted contour lines will generate areas where flat spots or steps are visible due to the geometry of the underlying TIN (triangular irregular network) used to determine slopes.
Hope this will help!
James Harvie
Vertical Mapper Product Manager
Northwood Geosience Ltd.
43 Auriga Drive, K2E 7Y8
Nepean, Ont, Canada
613-224-2020
