A Pocket Offline Map for the Kapadokya Valleys

Last April I spent six days walking the valleys between Göreme and Çavuşin with a phone that had no data plan and a 2016-era 32 GB microSD card. The usual offline map apps wanted 1.2–2 GB for the region at zoom 16, which was fine for storage but miserable for battery: every pan and zoom re-rendered vector tiles from the tile cache and the phone got hot enough to throttle by early afternoon. I wanted something I could leave on in my pocket, pull out at a junction, and read in two seconds without waiting for a map to think.

I ended up pre-rendering a fixed set of PNG tiles at zoom 15 and 16 for a bounding box roughly 38.60–38.72 N, 34.78–34.92 E, then stitching them into a single GeoTIFF and cutting it into 2048 px JPEG tiles at quality 82. The command chain I used, with gdal and ImageMagick on a laptop:

gdal2tiles.py --zoom=15-16 --tilesize=512 --webviewer=none \
  --profile=mercator valleys.vrt ./tiles_out

# then flatten: one image per ~2 km cell, 2048 px wide
for z in 15 16; do
  for x in $(seq 20860 20872); do
    for y in $(seq 13290 13305); do
      montage tiles_out/$z/$x/*.png -tile 4x4 -geometry +0+0 \
        -quality 82 flat_${z}_${x}_${y}.jpg
    done
  done
done

The flat cells totalled 640 MB, which fit on the card with room for photos. On the phone I used a tiny viewer that just opens the JPEG whose filename matches the current GPS cell, no tile fetching, no network. Cold open to a readable map was under half a second, and the phone stayed cool. The tradeoff is obvious: fixed zoom means no detail below about 1:8000, so I could not read individual rock-cut church outlines, only the paths leading to them. I carried a paper 1:25000 for the two valleys where that mattered.

Where this fails: the JPEG cells are mercator rectangles, so near the edges of the bounding box you get partial cells and the viewer shows white margins; I cropped the last row and column by hand. Also, if you walk past the box the viewer silently shows the nearest cell, which looks like a map but is wrong by up to 800 m. I lost twenty minutes above Çavuşin believing a cell that was actually the previous valley. Next time I would name cells by their southwest corner and show the coordinate in a corner of the screen.

What I am still unsure about: whether zoom 15 alone would have been enough for the valley paths, saving half the card. If you have walked these valleys with a fixed-zoom map, which zoom did you actually use, and did you ever regret not having vector tiles?