I carried out a geophysical survey identifying surface anomalies (fractures). The data was acquired through several lines of orientation profile W-E and N-S
Due to the topography of the terrain the profile lines are not equidistant (50m-200m) and have different orientations (W-E and N-S are preferential directions). However, the distance between measurement points of each profile is uniform (10m). In summary I have scattered data but with linear tendencies pertaining to the orientation of each profile.
In summary, I want to do an interpolation that takes into account two preferential directions (North-South and Norwest-Southeast interpolation) to generate a trend map within a GIS. An example is in Figure 1.
For the interpolation I have used the ArcGis and Surfer with several interpolation methods (Kriging, IDW, next neighboor, etc) but they do not produce good results because they interpolate data in unwanted directions (Figure 2)
I would like to know which method to use and which parameters I have to enter to perform the desired interpolation.
Thank you very much for your comments
Hey Carlos Armando Schwarck , I'm imagining that you're using the spatial analyst toolset to perform your spatial interpolation? I would recommend opening up the geostatistical analyst extension and attempt IDW or kriging. There are more parameters in this spatial interpolation toolset to better create what you're likely after.
Hi Carlos,
The main rule for the interpolation is the scattered points that covers the studied area such an spider web (Voronoi type). As in your case, there are maximum three points in each direction. So, in my opinion, you may not get satisfactory results from your data as the points are not dense enough and randomly distributed. For an alternative approach, you may use MATLAB interpolation methods (it is simple to apply, you can find examples in the web) such as cubic or v4 instead of linear interpolation...
Interpolation in GIS softwares draws a possible trend of the values but not actually create values to fill the gaps. So error will be there whenever you will do interpolation.
A random scattered dataset will generate a more aesthetically appealing map than then these lines.
For anyone to give you really useful information you need to be more specific about your data and why you want to interpolate. According to the graph/plot you have made measurements on some characteristic at each of the "dots"but it is not clear what the characteristic is. With respect to interpolation does this mean you want to generate more curves of dots or does it mean you want to generate values at some prescribed non-data points. What will you use the "interpolated data" for? It is premature to simply recommend that you use spatial analyst or specific algorithms such as IDW or kriging. Although there is no theoretical basis for IDW using it still makes some presumptions about the nature of the data, i.e you have a set of data locations and at each you have a measurement or recorded value, the objective is to generate a value at some pre-selected non-data locations.
Several respondents say or imply that you have to use GIS or GIS software, that is not true. Munir's comment about "drawing a trend" is wrong and also irrelevant but he is correct that there are always interpolation errors. As for kriging, in general it is better than IDW but it is based on various statistical assumptions and there are several different forms of kriging (Simple, Ordinary, Universal, Cokriging). If you are not already familiar with it, you should not just treat the software as a black box. The documentation is not enough to understand how to use it nor how to interpret the results.
Do a search on Google for "variograms and fractures"
The spacing between your lines of data points dictates the size of your search neighborhood and grid cell size, effectively making that search neighborhood larger if the points in a line are to influence the values between lines.
I tend to agree with Donald. Not enough information has been given to adequately determine the methodology to use. An understanding or the attributes of each data point is required, e.g., x,y,z coords, & whatever geophysical attribute or values. Your interpolation lines can then be determined on the basis of endpoint coordinates allowing you through intersecting lines to determine the profile you want along the interpolation line. This is easily done outside GIS software.
- Badges
- Science topic
- Similar topics
- Geoscience
- Cartography
- Mapping