stephenhann Posted March 7, 2016 Report Share Posted March 7, 2016 Hi, For a research project I have collected a square grid of soil samples along with a GPS co-ordinate at each location, I have tested the soils to find results such as pH, Organic Matter Content and the likes. I then took random samples within the Grid to test the accuracy of GIS when predicting soil characteristics. I am wondering which GIS method to use , i.e Interpolation, Kriging , Co-Kriging or any other method. The software I have available is ArcGIS 10.3. I was going to compare the results of the actual random samples compared to what the GIS predicted values would give. Thanks, Stephen Quote Link to comment Share on other sites More sharing options...
pasfans01 Posted March 7, 2016 Report Share Posted March 7, 2016 Prior to determine the best interpolation method, you should reconsider significant factors that determine variation of soil properties, like geology, elevation, slope, aspect, fluvial deposit zone, colluvial deposit zone, etc, from landscape ecology and geomorphology point of view, the distribution of soil properties heavily affected by those physical factors, so for example it will be pointless to doing interpolation of soil properties when the study area itself consist of mixed terrain or different geologic formation Quote Link to comment Share on other sites More sharing options...
stephenhann Posted March 7, 2016 Author Report Share Posted March 7, 2016 Yes I am aware of the Jenny model of soils forming factors, these have been taken into consideration. My project is just seeing how well the interpolation method works for a single site within a grid. Is there a preference of interpolation method depending on the physical properties? Quote Link to comment Share on other sites More sharing options...
quijote Posted March 7, 2016 Report Share Posted March 7, 2016 Hi Stephen Try regression/universal kriging to add auxiliary variables. I'm not sure ArcGIS 10.3 can do this, but I know gstat in R can do it! HTH! Quote Link to comment Share on other sites More sharing options...
spazzle Posted March 7, 2016 Report Share Posted March 7, 2016 Hi stephenhann There is plenty info available regarding which interpolation method to use, at the end of the day its what best fits your data in a meaningful way (KISS). It could be argued by some to use kriging, as this is its sole purpose is for use with soils etc. Give all the interpolation methods a try as time permits, starting with the basic and work up. There are specialised kriging tools which may be of interest, such as :- GAMMA, hXXps://www.gammadesign.com/ ISATIS hXXp://www.geovariances.com/en A version is available in CAE For GIS (The gisarea site) EVS (EVS uses expert knowledge to get the best possible results and can beat any proffessional at the kriging game) hXXp://www.ctech.com/ Good luck Quote Link to comment Share on other sites More sharing options...
pasfans01 Posted March 8, 2016 Report Share Posted March 8, 2016 yes there are some of them like all that already mentioned above, but as we are already know, interpolation is a form of unsampled/unknown value prediction technique, so either it is deterministic or stochastic/statistical assumption, it is hard to make a judgment which method is the best, because the phenomenon itself usually complex (interdependent each other, sometimes in unknown way),So my advice is, you better trying out some of interpolation method and compare the results with your known value samples, and conclude which is the best among them, your result usually couldnt be applied to other sites because the complexity of natural phenomenas 1 Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.