I'd be curious if someone who knows about video compression, someone with credibility, could explain the cause of these dark lines.
I do have some knowledge in the field of image manipulation and DCT compression, though I have been mostly self-taught in these areas so take my words with a grain of salt.
I agree with femr2 in that it's unlikely to be due to video compression artifacts; they get indeed worse near edges with high contrast but do not typically show the observed effect. Grainy noise in the 8x8 squares that the image is divided into is much more likely than pure vertical lines.
As an example, here's a zoomed in fragment of the above image:
(I've highlighted the 8x8 tile border lines. These are characteristic of JPEG but I think they apply to MPEG as well.)
Observe how there's a noisy zone in the edge but with no resemblance to the effect that we see in the building.
I don't agree with femr2's assessment that it's likely to come from the signal undershooting then overshooting. While the idea was appealing and I thought of it before I read his response, I think that if that were the cause, in the right edge of the building we would actually see first an overshooting (white line) then an undershooting (darker line than the sky). The scan wouldn't anticipate the future and undershoot first as happens when analyzing the signal as a whole (as in
http://cnx.org/content/m0041/latest/fourier4.png), but rather something more like this:
http://www.ganssle.com/articles/clip_image008.jpg
A third and more likely possibility in my opinion is that digital zoom was applied to the image. Some nonlinear interpolating algorithms used in digital zoom will exhibit such behavior, including Catmull-Rom and Lanczos.
Consider this test image consisting of a 25% grey rectangle inside a 75% grey background:
And look to what happens to the borders of the rectangle when zoomed 777% with Lanczos interpolation (cropped because it's a 995x995px image):
The digital zoom artifact hypothesis has some support in the fact that there are close to no features in the image with a thickness less than several pixels. The image is very lacking in high frequency information and that's likely because of digital zoom.
I'm not sure though. This image also shows dark/light lines, for example:
http://www.indymedia.org.uk/images/2006/03/336542.jpg
Yet there is a strong perspective effect that makes me think it's not digitally zoomed.
My second best guess is that it's due to applying some kind of sharpening to the image, but I highly doubt so.
I doubt that we can get a definitive answer without any further knowledge on the equipment and conditions used to film.