Mapping Density
- Density mapping shows where things cluster instead of where each single feature sits hence it is good for patterns, bad for pinpointing.
- Matters most when your polygons are different sizes; a raw count map makes a big polygon look busy just because it is big.
- Two ways you can go about it:
- Area method: We divide features by polygon area, or use a dot map. The dots are placed randomly, so they are a picture of density, not real locations. That feels like a trap for anyone who does not read the fine print.
- Density method: A raster where every cell gets a value from the features inside a search radius.
- Small cells give a smoother surface but cost processing time. A bigger radius smears the pattern out.
- The simple method just counts what is in the radius; the weighted method leans toward features near the cell centre and gives a cleaner map.
- Display with graduated colours or contour lines.
So we have these classificational schema; Natural Breaks, Quantile, Equal Interval, Standard Deviation which basically decides what the map tells, which is a lot of power for one dropdown. I keep wondering how often public data gets quietly skewed by someone picking an arbitrary search radius. Still, density fixing the unequalnpolygon problem is the real win here.
Finding What’s Inside
- Mainly for monitoring or comparing; some examples include; drug arrests near a school, or which zip code has more of something.
- Three methods:
- Area > Features: More visual. Fast glance, No data out of it.
- Select Features: Gives us a subset we can actually use for lists and summary stats (count, frequency, sum, average).
- Overlays: Merges boundaries and features into a new layer and permanently tags features with the area’s attributes. Vector is precise but leaves slivers; raster counts cells, faster but cell size drives everything.
To be honest Vector still wins for anything legal like parcel boundaries, because “close enough” does not hold up in a property dispute. However, for non high stakes thing the latter should be good enough.
Finding What’s Nearby
- In this chapter you ask some other different questions such as who is affected by an event, who is actually served by a facility and so on.
- “Near” is not only physical distance. It can be time, money, or effort, which changes the decision process.
- Three methods:
- Straight line: Buffers, Select within distance, or a continuous distance surface. Planar for a flat plane, geodesic for a curved earth. Ranges can be inclusive rings (0–1, 0–2, 0–3 mi) or distinct bands (0–1, 1–2, 2–3 mi).
- Cost over a network: Streets and other fixed infrastructure.
- Cost over a surface: Overland travel with no roads.
The RINGS vs BANDS distinction is the most useful thing in this chapter for me. Bands isolate each ring, so if I ever compare demographics by distance from a facility, that is the one I want. Building turntables from scratch sounds genuinely tedious, but for emergency routing I do not see how you skip it.