Alternative #2: Mapzen

The Mapzen dataset is an amalgamation of multiple source datasets, so is similar in concept to what I’m hoping to achieve with gpxz.

The dataset is seamless and voidless so is easy to work with. But it has some issues that I’d like to improve on:

  • Much of the world is based on SRTM, so inherits many of it’s issues.
  • I have seen some specific issues like noise in Chicago and large discrepancies in northern Scandinavia. Having a rolling-release dataset would mean I could get feedback from users and address issues in a following release.
  • The dataset was last updated in 2018. Many high-resolution datasets have been released or updated since then, and given the status of Mapzen is unlikely to be updated again.
  • The dataset is sampled at a 1 arc-second resolution
    • This is much lower than some of the source datasets, reducing precision.
    • High-resolution datasets are often released in UTM or an equal-area projection: reprojecting to EPSG:4326 further reduces accurately.
  • There’s not a lot of information about how the dataset was assembled.