So I have this raster dataset OS Terrain 50 which is 153MB, zipped, on disk.

I import a subset of these ASC into Manifold, importing about 50 of these 10km x 10km tiles, each around 191KB on disk. I then merge these tiles into a single image.

I end up with an image size of 140,000 x 160,000 (this is the rectangle of my AOI but I don't need all the data within that rectangle, 386,103 records, and my Manifold project balloons from about 3GB to 36.5 GB.

Am I missing something here? This seems disproportionate. Am I doing something wrong? I don't see how a small subset of a zipped 153MB dataset can balloon by 30GB. Appreciate that I might have crated some artifacts during the merge, and that potentially storage costs in the sparse matrix are not nill, but this is way beyond what I was expecting.

Is there a more efficient way to undertake what I am trying to do here - get a single, irregularly shaped large, tiled image by combining smaller images?

Using Manifold 9.0.173.2

The problem seems to be, when merging adjacent images, the tile count goes from 4 x 2 to 12403. Resampling being done somewhere. All the tiles are in the same projection.

Are using source metrics, or default metrics (1 unit per pixel)?

You might want to check your Merge workflow.

That would explain it, thanks. Must RTFM.

It's not necessarily obvious.

See this thread.

Shortcut: right-click one of the images in the Merge list, and select Use Coordinate System.