Scene from above

I’ve been severely neglecting my blog on account of focusing on writing up my PhD project as well as being sick (don’t underestimate the pain of getting your tonsils out as an adult!).

I wanted to write up a decent post for my 100th entry, but have subsequently realised it’s lead to me posting nothing for the last couple of months! I have a plan for a good entry coming up, though will need to find the time to put it together.

In the meantime, I picked up that Alistair Graham (geoger), who gave a talk at the conference I ran this year, and Andrew Cutts, who I have never met, though I remember worked through the straightforward openCV GUI demo from his website which I thought was great, have started a podcast, scene from above.

Science communication is tricky at the best of times, so I’m excited they’re giving this style of delivery a crack. The demo episode discusses Sentinel 5p and the larger scope of the sentinel project, remap’s webapp and cloud computing more generally, and the launch of a Moroccan satellite.

I think the discussion of the webapp was my favorite part. I appreciated Alistair’s humility in admitting that maybe he was approaching interaction with data from a point of view that was somewhat outdated, as he seems (as am I!) skeptical of the benefits of a sleek interface. Admittedly the app isn’t designed with me or others in the RS community in mind, but I can’t see it being used much in it’s current iteration.

Thinking of my ornithologist friends currently in PhDs/postdocs who would be the target audience for an app like this, they would almost definitely look at it for an hour or two with interest, and never think to use it again. Having consistently tried to get them interested in RS and accurate mapping, the tools need to be unbelievably simple to get people to consider using them seeing as so much of other scientists time is dedicated to learning specialist knowledge and general computing skills. It’s one of the many challenges of interdisciplinary work in science!

I’m looking forward to the next episode of the podcast, and hope a forum opens up for discussion online as I think I’d have something to contribute, and would love to hear other people’s opinions on these ideas!

Keep an eye out for a longer update soon 🙂


Choosing a field site: The hunt (Part three)

In another portion of the same field trip, some students were required to do a survey around the town of Hunstanton, about an hour’s drive from the earlier stop of Overstrand. Hunstanton is regionally known for its chalk cliffs, which my supervisor had previously suspected might be a good site candidate. So, after dropping off the students, we went down to investigate its potential as a study site.

The aerial image (once again courtesy of GetMapping) reveals a sharper cliff, far more distinct than what appeared to be a set of landslides at Overstrand. Looking within the shadows reveals numerous rockfalls and slips, which was thought could make for an interesting (and challenging!) environment for photogrammetric mapping. The tide in the image is almost at its highest, a search reveals that it shouldn’t be a big issue for conducting a survey as it stays relatively far out, but this will obviously be influenced by focal length and sensor size of the cameras we use!


Aerial image of Hunstanton and its cliffs, courtesy of GetMapping

The Hunstanton cliffs are extremely well documented, and another set of cliffs with conservation status. One of the earliest records revealed is an 18th Century essay, ‘Towards a topographical history of the county of Norfolk’, with numerous mentions of the Hunstanton cliffs. The cliffs consist of three distinct strata denoted by their striking colour differences (colour contrast!). The bottommost layer is rusty brown Carstone, which contrasts with the striking Hunstanton Red Chalk Formation and white/grey topmost Ferriby Chalk Formation. It’s a popular area with fossil hunters, being a very productive area for ammonite fossils.


Strata at Hunstanton, taken from the GCR site description

This definitely ticked the colour contrast box in a big way, and appeared to be very accessible – there’s a carpark located at the top of the cliff allowing an easy path to the strand. The reconnaissance mission revealed one very interesting 50 m or so stretch where a landslip had occurred, according to a passer-by, in the early nineties.


An overview of the cliffs of Hunstanton, including the aforementioned landslip

Vegetation has taken hold in upper parts of the landslip to create a natural scene with an almost unnatural amount of colour contrast, things were shaping up nicely! One of the only downsides we foresaw was a large amount of traffic through the area, but this wasn’t a large enough of a deterrent to add it to the list.

This set the stage nicely – two sites within an hour of each other, with good options for accommodation and well accessible. Now all that was required was careful planning and enough time to ensure the whole thing was done correctly! A checklist and research design was made of the most important equipment needed to answer the questions posed. I accessed the UK LiDAR inventory in order to make some coarse resolution 3D models of the areas of interest, and used a nifty CloudCompare function to actually place virtual cameras in the scene. This came in useful for planning what focal length lens would be required for data acquisition!


Overview of the initial plan for the photogrammetric survey at Hunstanton, showing camera positions

I actually spent some (probably too much!) time fiddling around with a Blender plugin which simulates a physical camera, so you can play with the focus distance/aperture, but nothing much came from it. Something on the back-burner.

Stay tuned for part four, where I detail the fieldwork plan.