Search this site

Navigation
Home‎ > ‎

Overview of Our Project, HERO MAP



We are the HERO MAP team.  We work at Clark University in Worcester MA as part of a Human Environment Research Observatory.  We are affiliated with the PIE LTER, and are NSF funded.  We make .5m resolution land-cover maps using free 45cm aerial photographs from MassGIS, as well as ancillary data from MassGIS and the towns that we classify.

Here's how we do it:

  1. Acquire data for one town
  2. clean up data, merge as necessary
  3. split into girds of manageable file sizes
  4. segment to make image objects
  5. load or build a classification hierarchy
  6. classify
  7. manual edits
  8. merge image objects by land-cover class
  9. manual edits
  10. export as raster
  11. convert classification to vector
  12. merge grids into whole town
  13. match edges of grids to fit better
  14. create strata for validation
  15. validate through virtual field work
  16. create soft classification
  17. determine mean NDVI by parcel for fine green areas
  18. analyze land-cover proportions for the whole town by parcel, census block, and town.  Make tables of these proportions

The rest of the tutorial is about understanding the specifics of these steps, and how to get around problems that may appear.