In this challenge you're given data about some of Melbourne's Statistical Areas (as defined by the ASGS - Australian Statistical Geography Standard) and about the City of Melbourne's urban forest. Statistical Areas are given in the form of polygons representing their borders. The urban forest is given in the form of polygons too, where they represent the area occupied by tree canopies or bush. In both cases, polygon vertices are longitude/latitude pairs.
More information about Statistical Areas can be found at the ASGS web site: http://www.abs.gov.au/websitedbs/D3310114.nsf/home/Australian+Statistical+Geography+Standard+(ASGS).
The challenge consists in determining the greenest suburb of Melbourne, where greenest is the suburb with the highest vegetation per area ratio.
Some relevant info here:
- Suburbs are related to Statistical Areas Level 2 (SA2s).
- File melb_inner_2016.json contains the Statistical Areas of inner Melbourne.
- Directory melb_urban_forest_2016.txt contains the urban forest of the City of Melbourne council.
- In both datasets coordinates are longitude/latitude pairs.
- You can check http://s2map.com to play around with plotting polygons on a real globe map (choose Statemen Terrain in the drop-down options). This is another good one for that: http://apps.headwallphotonics.com/
For this challenge, you will most probably want to use Spark/PySpark or Beam/Dataflow, as they provide the adequate parallel processing you'll need to get results in a reasonable amount of time. You can choose any language or environment you have access to.
Some useful types and functions to help with processing polygons are also provided. Type Polygon is defined according to the GeoJson definition of polygons, i.e. a sequence of loops. In a sequence of loops, the first loop (index zero) is the polygon's outer border (perimeter), while any following loop (if present) represents holes within the outer border. Type Loop is a sequence of points and Point is a sequence of 2 doubles. There is yet another type called MultiPolygon, which is simply a collection of polygons. Multi-polygons are very useful to represent disjoint areas that make up the abstraction of a single piece of territory, for example, an island with the piece of coast to which it belongs.
Using jar challenge-urban-forest-jar-assembly-1.0.0.jar, import au.com.eliiza.urbanforest._ and those types will be
provided like:
type Point = Seq[Double]
type Loop = Seq[Point]
type Polygon = Seq[Loop] // where the first loop is the outer border and the following ones are holes
type MultiPolygon = Seq[Polygon]
Four functions to process multi-polygons are provided:
// calculates the area of a multi-polygon
def multiPolygonArea(multiPolygon: MultiPolygon): Double
// calculates the intersection area of 2 multi-polygons
def intersectionArea(multiPolygonA: MultiPolygon, multiPolygonB: MultiPolygon): Double
// merges n multi-polygons
def mergeMultiPolygons(multiPolygons: MultiPolygon*): MultiPolygon
// a cheap way to check whether 2 multi-polygons might intersect (based on their bounding boxes)
// NOTE: if the return is false, the multi-polygons are GUARANTEED to NOT intersect; however,
// if the return is true, the intersection is not guaranteed
def mayIntersect(multiPolygonA: MultiPolygon, multiPolygonB: MultiPolygon): Boolean
Equivalent functions are given in Python too! Just import everything from polygon_utils.py.
Using the same jar challenge-urban-forest-jar-assembly-1.0.0.jar, import class au.com.eliiza.urbanforest.j and
everything in it (au.com.eliiza.urbanforest.j.*) to gain access to equivalent types and methods.
Good luck!!!