I love models

GIS models, that is…


I get to do fun things with GIS, like develop models that handle and process millions of records.  This model is used to clean a number of spatial variables for input into another analytical model.  This awesome monstrosity contains:

  • 205 million input xy points read from SQL tables that are processed into over 500,000 polygons
  • 500,000 polygons merged and manipulated into about 200,000 polygons
  • Over 20 different spatial processes (merge, clip, field calculations, projections, spatial stats, etc…)
  • Database driven model, using a mix of SQL tables, SDE datasets, and File Geodatabases
  • Four custom python scripts, some of which are repeated several times
  • The output? Two text files, which are generated from python scripts.
  • Output used in another set of spatial models

The model took about a week to build and calibrate.  The model itself runs in about 30 hours on a four core, 64 bit machine and produces close 20GBs of data from about 4GBs of input data.

Anyone can build a mapping website (including me!), but a true GIScience geek lives on this stuff.  I love spatial analysis.