It got a bit quiet on this blog lately. But not without reason. For the last 6 months, we were hard at work with analysing and publishing the results of the Borre Monitoring Project (BMP) – a pilot study we undertook a few years before the start of VEMOP.
BMP emanated from a somewhat worrisome observation made in July 2015, when NIKU surveyed an area containing two Iron Age hall buildings at the outskirts of the Borre National Park in Vestfold using their newly acquired MALÅ MIRA system. The halls had been discovered in 2007 and since then had been surveyed again in 2008 and 2013. Each survey confirmed the presence of two large buildings visible through patterns of postholes. The survey in 2015, however, failed to do so.
During data analysis in 2016, we looked at potential reasons for this unexpected outcome, with the culprit quickly identified: unusual amounts of rainfall in the week prior to the 2015 survey, leading the subsurface to be close to saturation during data acquisition.
Following this realisation, we decided to look more systematically into the issue at Borre. In the summer of 2016, we were able to rent the area covering the hall buildings for a period of 14 months from the landowner. This allowed us not only to conduct repeated GPR surveys over the course of a year in different conditions (read: wet, dry, cold, etc.), but to directly measure volumetric water content, bulk electric conductivity and ground temperature of the subsurface during each survey and to compare them. The aim was to better understand how environmental factors influence the quality of our GPR data, or simply put: How much rain is too much?
In general, the results showed that the quality of the GPR data sets collected during the monitoring period was mainly controlled by small-scale weather patterns such as local rainfall and temperature changes. Seasonality was not observed.
Grouping the data sets into low and high quality and looking at the conditions under which they had been collected, underlined the importance of dry conditions for GPR surveys. While this observation in itself was not overly surprising, the impact rainfall events could have even on a sandy, well-drained site such as Borre came unexpected. The reason for this, we think, can be found in the thick, organic-rich topsoil, which can store large amounts of soil moisture after rainfall events, which in turn leads to greater attenuation of the electromagnetic signal. If the contrast between the buried archaeological features and the surrounding subsurface materials beneath the topsoil is not that great to begin with, surveys can fail. This is likely what happened during the survey in 2015.
But the results also showed that not only rainfall intensity alone but the chronology of events played a role in how well the hall buildings were visible in the GPR data. More time between rainfall and survey allowed for more drainage and thus drier conditions, leading to a better quality.
Wintry conditions provided the most interesting and surprising results of the BMP. The best data set was collected under very cold temperatures and low humidity. The partially frozen topsoil allowed more electromagnetic energy to enter the ground, resulting in GPR data sets of higher quality. Interestingly, just a month later, in February, we collected the data set of the lowest quality, in which the hall building was almost invisible. A sudden change in air temperatures had triggered the topsoil to thaw and the previously favourable conditions quickly took a turn for the worse. Water puddles on the surface and an on-going thawing process in the topsoil lead to much of the electromagnetic signal being attenuated and reflected in the upper parts, rendering any existing contrast between postholes and surrounding subsurface materials in the layers beneath meaningless.
These observations point to winter as a possible extension of the traditional GPR field season before sawing and after harvest – if the right conditions prevail.
The data collected for the BMP provided a great opportunity to test different analytical ways in which to handle the large amounts of GPR and monitoring data we accumulated for VEMOP. One example for this is the on-going development of Schlitzi + which we discussed in one of the previous blogs.
If you are interested in all the details of the Borre Monitoring Project, have a look at our paper, it’s free 😊