02-22-2024 12:44 PM
Labels (3)
New Contributor
Čt 22.02.2024 21:35
Good morning,
I would like to ask for advice on the best interpolation method for interpolating 4,000 points onto a map covering a large area. Specifically, these points represent soil data, with each point serving as a sampling point providing various soil characterization data. The distribution of these points across the area is not uniform; hence, some parts have denser point coverage than others. My aim is to create a comprehensive map covering the entire Czech Republic and Slovakia.
Currently, I am employing Empirical Bayesian Kriging (EBK) with Empirical data transformation and utilizing the k-bessel semivariogram. However, I am uncertain whether this is the optimal choice or if there are better alternatives available. I would greatly appreciate any insights into selecting the most suitable interpolation method. Additionally, I am concerned about potential issues related to the size of the area. Since interpolation is based on straight-line distances, I am worried about inaccuracies in distance calculations. Is there a way to quantify and mitigate these inaccuracies? Furthermore, I am unsure whether ArcGIS Pro offers algorithms to address this issue.
Thank you for your assistance.
Tags (1)
0 Kudos
1 Reply
New Contributor II


There is a post from Esri if you want to compare the capabilities of each interpolation method here ( yet it doesn't say the trade-off of each method. You may want to read some journals to find the pros and cons of each method since your case might be very specific on the topic of soil things. For example, you can read in this paper (

Another way to quantify the accuracy of each interpolation method is by using the Cross-Validation Tool in ArcGIS Pro ( For example, I created two interpolations, IDW and Empirical Bayesian Kriging (EBK) (make sure you create using Geostatistical Analyst Tool one). I can compare the accuracy of both interpolations from its error value (choose the smaller error). 


The result from this tool will create a new point feature class containing the actual and predicted values, followed by its metrics. For example you can sum the error column to get the sum error metrics. Then compare it with another interpolation method


Hopefully answer your question

0 Kudos