Mexico 5 meter DEM

[nvkelso]

• http://www3.inegi.org.mx/sistemas/productos/default.aspx?cl=209010200
• http://buscador.inegi.org.mx//search?q=&requiredfields=cve_titgen:3864&client=ProductosR&proxystylesheet=ProductosR&num=10&getfields=*&sort=meta:edicion:D:E:::D&entsp=a__inegi_politica_p72&lr=lang_es%7Clang_en&oe=UTF-8&ie=UTF-8&entqr=3&filter=0&site=ProductosBuscador&tlen=260

[iandees]

I've spent a couple hours trying to find the CRS for this data and haven't found anything that seems to work. Based on the "DATUM" attribute on an entry like this one, I've been looking for information on ITRF08. There's entries like this one in epsg.io, but they don't produce an image in the right place on earth. I'm still looking for the right projection.

Once I do find the right projection, there are a couple changes I need to do to make the transcode step work:

• The environment variables need to get updated to include the .xyz file type
• The transcode.sh script needs to get updated to find a .xyz file inside a .zip file

[iandees]

Did a bunch of pre-processing (similar to the UK data) documented in this new repo: https://github.com/iandees/mx-lidar

Transcode files

aws s3 ls s3://elevation-sources-prod/mx_lidar/ --recursive | \
grep .tif$ | \
awk '{print $4}' | \
while read filename; do \
    bn=$(basename ${filename%.*}); \
    make submit-job job=aws/transcode-job.json.hbs input=s3://elevation-sources-prod/${filename} output=s3://elevation-sources-transcoded/$(dirname $filename)/${bn} name=${bn:0:50}
done

[iandees]

The transcoding finished and we ended up with 2,516 footprints in the database. This doesn't come close to covering Mexico, though:

This regex filter in the URL listing step cut out most of the ~50k images from the portal. I think the rest of the images are "surface" (i.e. with buildings) images?

The data that did make it in looks pretty nice, though:

[iandees]

I've poked around a bit on other pieces of the data at INEGI this morning and there's more to pull in, but almost none of it has valid projection information in a machine-readable format. I'm going to call it quits for now, as the next step (digging through 50,000 URLs to find a pattern in projection information) is vastly more time-consuming.

[nvkelso]

@iandees I am seeing Mexico data in the dev tiles (which looks good). Can you confirm this issue was completed in the v1.1 milestone, please?

Eg around here: http://localhost:8000/#13/22.7620/-102.5455

• http://localhost:8082/normal/13/1762/3564.png x-amz-meta-x-imagery-sources:mx_lidar/889463090595_b.tif, srtm/N22W103.tif

This area south of Mexico City also looks much better with the new Mexico data:

• http://tangrams.github.io/walkabout-style/#13/18.7046/-98.3015

[iandees]

Yep, there are 2,503 footprints in the database for mx_lidar (this issue), but as mentioned in the comment above it covers a very small part of Mexico.