How we compare more than 100 GPR data sets

Data visualisation and analysis in Schlitzi+. All GPR data from one test site are accessible (upper right corner), creation of arbitrary sections through each processed dataset at freely definable locations (lower left corner), extraction of single GPR traces from the collected raw data for direct comparison (lower right and upper left corner). Foto: Erich Nau/NIKU




10.02.2023 kl.11:34

Throughout the past one and a half years, we collected more than 100 individual GPR data sets for VEMOP by surveying roughly once a month with two different systems at four test sites. A huge amount of data and a challenge, when it comes to the actual data analysis.

When dealing with conventional GPR surveys, both NIKU and Vestfold and Telemark Fylkeskommune use the software ApRadar, developed by the Austrian Ludwig Boltzmann Institute for Archaeological Prospection and Virtual Archaeology, as the main tool for data processing. ApRadar has been specifically designed for the use in archaeological prospection as opposed to programs that mainly aim for the geotechnical and utility markets. It transforms the raw data by applying advanced processing and visualization algorithms into georeferenced depth-slices needed for subsequent archaeological interpretation.

However, while we are big fans of ApRadar and value the opportunity to be able to personally contact its developer Alois Hinterleitner whenever we have a particularly tricky data set, our conventional way of looking at GPR data (aka depth-slices) is not detailed enough for the questions we try to answer within VEMOP. To compare our data sets and better understand how environmental factors do influence the contrast in our GPR data sets, we need to look deeper: at the raw data and single traces, individual radargrams as well as computer generated arbitrary profiles.

During the Christmas holidays in 2020, Erich started to play around with “Karel the robot”, a free and simple teaching environment for programming basics. After mastering Karel, he decided to continue to improve his coding skills and apply them to his actual work. Several hundred hours of learning the basics of the python programming language (mainly using trial and error) as well as searching Stack Overflow and GitHub later, Erich has developed the Schlitzi+ toolbox, a python-based analysis tool that enables us to quickly and easily access the full scope of our GPR datasets.


skjermbilde fra et dataprogram
Input window used to organize all single monitoring surveys in a database within Schlitzi+. Foto: Erich Nau/NIKU


Schlitzi+ creates a project-specific database, storing and organizing all datasets from a single monitoring site: Name, survey date and -time, GPR system as well as links to the raw data, intermediate processing steps and final depth-slice images.

An image viewer allows to read the database file and to interactively access all data stored in it.  That way, we can easily scroll through the depth-slice images and switch between all of the GPR datasets acquired at each test site. Basic GIS functionality allows to select points and lines based on the depth-slices and thus easily retrieve individual traces from the raw data or generate arbitrary profile throughout the 3D data block. This can subsequently be used for comparing the GPR data sets at one test site to each other, as well as to the results of the in-situ soil moisture sensors and the weather stations -  a task that is crucial for VEMOP

The Schlitzi+ toolbox is currently still under development, but already it enables us to study the collected GPR data sets in greater detail and a more convenient way than previously possible. More about that in the next blog.


Geofysikk Arkeologi