Simulating feature detection by trial trenches and its consequences for site detection and evaluation

Description

Trial trenching is the survey method of choice in many areas in Europe, being capable of covering substantial areas while only sampling between 5 and 10% of the surface. However, previous research (Hey and Lacey 2001; Verhagen and Borsboom 2009) has also shown that the detection and interpretation of features uncovered by trial trenching may be seriously influenced by the survey strategy chosen. In particular trench length, width, spacing and configuration determine the results of the survey. While the effects of manipulating these parameters are quite well understood in statistical terms, empirical data to quantify the risks of under- or overestimation of feature densities has largely been lacking so far.

Flanders Heritage Agency therefore has carried out simulations of various trial trenching strategies on excavated site plans from Flanders in order to better understand these effects. The results of this study clearly indicate that low area coverage results in a higher risk of feature densities being wrongly estimated, and thus confirm earlier research results based on statistical modelling. Secondly, the simulations underline the large inherent variation in the possible results of trial trenching campaigns.

The method employed can easily be transferred to other site types and survey methods to estimate the risk of survey results deviating from the actual feature density. It is therefore a useful tool to find the balance between the research effort applied and the reliability of survey results.

Period	28 Oct 2016
Event title	Workshop "(Re)thinking archaeological potential in preventive archaeology"
Event type	Workshop
Degree of Recognition	International