One of the most important decisions within a PhD involving fieldwork is the correct selection of a study site. Indeed, there are many long-term studies where the study site is static and all the fieldwork revolves around it (The Supersauze landslide, or any of the Smithsonian field sites spring to mind).
Given the questions I have been posing throughout my PhD about how image quality and network geometries can effect photogrammetric accuracies, it was decided a field campaign should be run in an attempt to properly address them. This lead to thinking about what an ideal fieldsite would look like and broadly the properties it should have. I split these originally into two camps:
With all the gear we had planned to bring in, this was top of the list. The closer we could get to the field site, the more we could practically achieve within the time frame
More in environmental sense, given the sensitive nature of using equipment on tripods, one of the main goals was to have everything static for the duration of each survey
We needed an area from where we could head out early in the morning and be within reasonable distance of the target site
Carrying out a photogrammetric survey with a crowd moving through the scene has obvious drawbacks. While this conflicts with accessibility (As easier access usually means more people), it was a consideration nonetheless
More to do with the timing, though access to seafronts where tides could potentially have an effect were a consideration
Correct scale + conditions
One of the main reasons to do fieldwork as opposed to performing everything in lab conditions, is that we can practically get an intuition for how the technology is performing in the field. This includes having to deal with rain/wind/exposure differences in practice
For my blog’s followers, you’ll know I’ve spoken about colour contrast before. One of the prevailing themes to be tested, we wanted a scene with ideally vegetation, mud/sand and exposed rock
Given a UAV was not to be used for this work, we were looking for a scene we could capture images from the direction of the surface normal (Perpendicular to surface).
Thus, cliffs were a natural consideration
Refining the selection
Taking account of all the criteria, I took to the ordnance survey maps to try and see the options available. For a moment I thought about how to phrase it as a database query, I think it’d be a pretty nifty query to be able to do, but practically difficult. Something like ‘I want a 100 m stretch of cliff which is partially vegetated and easily accessible within 1 hours drive of a town with a hotel’. For fun, I thought I’d check if google was up to the task.
It lead me to a rambler’s guide to the Devon coast, and what looked like some seemingly suitable cliffs to use as experimental subjects.
Devon is quite a way from London, but was certainly within the bounds of possibility. In part two I’ll elaborate on thinking which sprung from this, leading to a final selection.