That really depends on your data. You're fairly limited in what you can do with categorical data which it sounds like your texture data might be.
You can do more with ratio/interval data, but you still need to examine the distribution and heterodasticity of the data (through data exploration tools like histogram analysis, semi-variograms, etc that are part of Geostatistical toolkits in programs like ESRI's ArcGIS desktop).
Now, density analysis like kernel density analysis is looking at the positions of the points rather than the attributes of the points, which isn't what you want. You can weight points using a population field, but the idea is to look at the locations of the points themselves - for example, density of crimes across a city. Were there 10 robberies in this neighborhood or 20? Etc. But in this case, all that analysis would be telling you is the distribution of your sampling points. Here's ESRI's page on kernel density analysis:
http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#//009z00000011000000.htm (note - even if you don't use ArcGIS, ESRI's help pages are good introductions).
If you want to create a surface based on the attributes of your data, which seems more likely, then you're talking interpolation. You have two types. You mention IDW. That''s one method, as is kriging...
So when do you use density analysis vs. interpolation? I think O'Sullivan and Unwin put it clearly in Geographic Information Analysis (p234). If it is THEORETICALLY possible to measure the value of an attribute anywhere that you haven't got a data point, use interpolation. In your case, you may only have 100 data points for soil salinity, but you COULD in theory take a measurement anywhere there was soil. That's very different than non-continuous data like crime counts or disease incidents...
So what TYPE of interpolation? IDW is deterministic. It uses a specific mathematical function to calculate the values between your data points and includes no measure of uncertainty. More complex surface analysis like kriging has the advantage that it produces error measurements - you can get a sense of how accurate your surface is by analyzing the residuals (errors) produced by the surface. Here's a good intro page with more detailed pages linking off of it: http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#//009z00000069000000.htm
I would start with IDW if I were you, just to get a sense of the data. Then consider kriging. If you decide to create kriging surfaces, make sure to take the time to do data exploration (semi-variogram, histogram etc) first to see if your data needs to be transformed or have trends removed. This tutorial: http://help.arcgis.com/en/arcgisdesktop/10.0/pdf/geostatistical-analyst-tutorial.pdf is a good way to familiarize yourself with the tools if you haven't used them before. Again, even if you use a different GIS, ESRI's help products are good introductions to the theory. I'd also recommend the book I mentioned above - David O'Sullivan and David Unwin's Geographic Information Analysis. It's a very readable system-independent introduction. Without knowing more about your data or objectives, that's about the limit of my suggestions, but I hope they've been helpful!
That really depends on your data. You're fairly limited in what you can do with categorical data which it sounds like your texture data might be.
You can do more with ratio/interval data, but you still need to examine the distribution and heterodasticity of the data (through data exploration tools like histogram analysis, semi-variograms, etc that are part of Geostatistical toolkits in programs like ESRI's ArcGIS desktop).
Now, density analysis like kernel density analysis is looking at the positions of the points rather than the attributes of the points, which isn't what you want. You can weight points using a population field, but the idea is to look at the locations of the points themselves - for example, density of crimes across a city. Were there 10 robberies in this neighborhood or 20? Etc. But in this case, all that analysis would be telling you is the distribution of your sampling points. Here's ESRI's page on kernel density analysis:
http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#//009z00000011000000.htm (note - even if you don't use ArcGIS, ESRI's help pages are good introductions).
If you want to create a surface based on the attributes of your data, which seems more likely, then you're talking interpolation. You have two types. You mention IDW. That''s one method, as is kriging...
So when do you use density analysis vs. interpolation? I think O'Sullivan and Unwin put it clearly in Geographic Information Analysis (p234). If it is THEORETICALLY possible to measure the value of an attribute anywhere that you haven't got a data point, use interpolation. In your case, you may only have 100 data points for soil salinity, but you COULD in theory take a measurement anywhere there was soil. That's very different than non-continuous data like crime counts or disease incidents...
So what TYPE of interpolation? IDW is deterministic. It uses a specific mathematical function to calculate the values between your data points and includes no measure of uncertainty. More complex surface analysis like kriging has the advantage that it produces error measurements - you can get a sense of how accurate your surface is by analyzing the residuals (errors) produced by the surface. Here's a good intro page with more detailed pages linking off of it: http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#//009z00000069000000.htm
I would start with IDW if I were you, just to get a sense of the data. Then consider kriging. If you decide to create kriging surfaces, make sure to take the time to do data exploration (semi-variogram, histogram etc) first to see if your data needs to be transformed or have trends removed. This tutorial: http://help.arcgis.com/en/arcgisdesktop/10.0/pdf/geostatistical-analyst-tutorial.pdf is a good way to familiarize yourself with the tools if you haven't used them before. Again, even if you use a different GIS, ESRI's help products are good introductions to the theory. I'd also recommend the book I mentioned above - David O'Sullivan and David Unwin's Geographic Information Analysis. It's a very readable system-independent introduction. Without knowing more about your data or objectives, that's about the limit of my suggestions, but I hope they've been helpful!
I recommend vertical mapper. Https://www.researchgate.net/publication/233841508_using_mapinfo_vertical_mapper_interpolation_techniques_for_estonian_oil_shale_reserve_calculations
Conference Paper USING MAPINFO VERTICAL MAPPER INTERPOLATION TECHNIQUES FOR E...
pH generally shows good correlations with terrain parameters such as altitude or slope. Try regression analysis is this case. If you have enough data, you can also test the geographically weighted regression, which is implemented in ArcGIS and SAGA-GIS. Check out the article: http://www.revagrois.ro/PDF/2009_1_417.pdf
I recommend you the use of geostatistical tools -estimation like kriging or simulation-.
For this type of applications (ph mapping), I prefer conditional simulation geostatistical techniques that preserve the variability of the original data (the conditional simulation preserve the min, max values, the mean and variance, the histogram and the spatial variability expressed as variogram fuctions).
You can apply them using free software tools like Gslib (http://www.gslib.com/) or the mGstat (a geostatistical Matlab toolbox) http://mgstat.sourceforge.net/.
I hope that this recommendation will be useful for you.
Ibeside Geostatistic, you can combine with satellite images to differenciate the land cover which correlated with soil properties, then combine with the one you intepreted for more accurate delineation the soil properties