Data & methodology ▾
Grid: 1,493 landscape-metric analysis cells across Dortmund's twelve Stadtbezirke, each assigned to one of five spatial regimes (Core, North, South, East, West City) — the same regimes used by the companion Spatial Durbin regime model.
Model: single-scale GWR (adaptive bisquare kernel), fit independently per metric and per year at that model's own optimal bandwidth — not a fixed bandwidth reused across years.
| Metric | 2015 | 2025 | 2035 |
|---|---|---|---|
| Built-up PD | 205 | 205 | 145 |
| Veg. Cohesion | 608 | 657 | 860 |
| Veg. PD | 253 | 194 | 209 |
Local R² is the model's goodness-of-fit at that specific cell. coef(X) is that cell's local regression coefficient for standardized covariate X — positive (blue) means X and the metric move together locally, negative (red) means they move oppositely.
Dashed outline flags a cell whose coefficient is more than 3 standard deviations from the layer's mean — a sign of local estimation instability (common at the edge of a bandwidth's neighbour set), not necessarily a substantive effect.
All nine models were refit live from the source shapefiles for this map, at each year's known optimal bandwidth — the values here are not copied from the static notebook figures.
Grid geometry reprojected to WGS84 and simplified for web delivery. See the full case study for the underlying coefficient-snapshot maps and cross-metric comparison.