LIDAR data from DEFRA

A blog post describing this work: http://www.ianhopkinson.org.uk/2015/10/analysing-lidar-data-for-the-uk/

Usage

process.py {OS_grid_cell} {friendly_name} will display a window showing the data from a LIDAR map tile. OS_grid_cell is, for example, SJ46 for Chester. The {friendly_name} will later be used in the display of menus and is optional.

The data is assumed to be in a zip file with name of the form:

LIDAR-DSM-2M-{OS_grid_cell}.zip

These data must be downloaded manually from:

http://environment.data.gov.uk/ds/survey#/download

So far I've successfully used the 2m resolution Terrain and Surface files.

Run ./process.py, and after a shortish pause (<1 minute) you should see a greyscale image of your map tile with height encoded as shade (black = lowest, white = highest). This image is written at full resolution to the images/ directory with the name {OS_grid_cell}.png.

Nominally there are 100 subtiles to a set but some are missing and appear as black squares, some locations have NODATA, these also appear black in the image.

This assumes you have numpy and matplotlib libraries installed.

There are minimal tests which can be run by doing nosetests, assuming you have nosetests installed

There are currently two experimental HTML/Javascript visualisations. To run them, run the minimal webserver in Python 3:

python -m http.server 8888 &

And then navigate to localhost:8888 for a leaflet.js map overlay visualisation, the available images and their bounding boxes will be picked up from the data_dict.json file which is updated when process.py is run.

Navigate to localhost:8888/surface.html to see a whizzy 3D surface rendering, just edit in the appropriate image file name to surface.html. (I've been making 512x512 pixel crops of the full sized images using Paint .NET.) Holding left-mouse (or A) and moving the mouse rotates the view, right-mouse-button (or D) pans the view, middle-mouse-button (or S) zooms the view.

TODO

For 25cm and 50cm resolution data don't render a whole 10kmx10km tile - render individual tiles or make a tight bounding box?
Trying to process 50cm dataset we get a memory low error, and then a MemoryError on trying to imshow, 20000x20000 pixel png successfully created though
Trying to process a 25cm dataset we get off-by-one errors in yoffset (and the image is too large for imshow)
Fix mismatch problem with the leaflet.js visualisation, could this be the problem:

https://help.openstreetmap.org/questions/2056/using-the-ordnance-survey-national-grid-with-openstreetmap

Or it might be that OS uses transverse Mercator and OSM uses spherical Mercator projection