All work
Modelling change · Multi-criteria site suitability

Weighing five layers to site solar farms

A transparent weighted-overlay model that scores every cell of the West Bank for how well it suits a solar farm, balancing terrain, sunlight, farmland value, and distance to towns, then distilling a noisy surface into a handful of clean candidate zones.

The five criteria behind the model: solar potential, distance to towns, slope, aspect, and agricultural value, each reclassified to a common 1–5 suitability scale across the West Bank.

The five criteria behind the model: solar potential, distance to towns, slope, aspect, and agricultural value, each reclassified to a common 1–5 suitability scale across the West Bank.

01The idea

Siting a solar farm is a negotiation between physical factors that rarely agree. You want gentle, south-facing slopes with strong sunlight, on land that isn't prime farmland and isn't stranded far from where the power is used. No single layer answers where should this go?. The answer lives in how you weigh them against one another.

So I built a small, transparent multi-criteria model to do exactly that, with the West Bank as the worked example. The point wasn't a one-off map; it was a method a planner could re-run, re-weight, and defend.

02The method

The model runs as a short, deliberately legible pipeline:

  • Five criteria, one scale. Aspect, slope, solar radiation, agricultural value, and distance to urban areas are each reclassified onto a common 1–5 suitability scale (5 best, 1 worst), and existing built-up areas are masked out entirely.
The five inputs, each reclassified to the 1–5 scale (5 = green, 1 = red). Slope and aspect come from terrain; agricultural value protects productive land; proximity keeps sites near demand.

The five inputs, each reclassified to the 1–5 scale (5 = green, 1 = red). Slope and aspect come from terrain; agricultural value protects productive land; proximity keeps sites near demand.

  • A weighted sum. The layers collapse into one continuous score per cell, with terrain carrying most of the weight: slope and aspect at 0.35 each, agricultural value 0.15, solar radiation and urban proximity 0.05 each. Aspect and slope dominate because a poorly-oriented or steep site undercuts everything else.
The weighted suitability surface: every cell scored from low (blue) to high (red) by combining the five layers.

The weighted suitability surface: every cell scored from low (blue) to high (red) by combining the five layers.

  • Group, denoise, cluster. The continuous score is grouped into four classes (very good, good, ok, avoid), then passed through a spatial filter to strip pixel-level noise, and clustered into contiguous zones that are exported as shapefiles.

03What it shows

The filtering is what makes the output usable. A raw suitability surface is too speckled to act on; grouped and denoised, it resolves into coherent candidate zones ranked by quality: the form a planner can actually take to a table.

From grouped classes (left) to the filtered, clustered result (right). Very good = green, good = blue, ok = yellow, avoid = red. The cleanest ground pulls away from the steep central highlands.

From grouped classes (left) to the filtered, clustered result (right). Very good = green, good = blue, ok = yellow, avoid = red. The cleanest ground pulls away from the steep central highlands.

And because the weighting is explicit, it's adjustable: favour flat ground over orientation, or protect more farmland, and the zones redraw. That makes it less a single verdict than an instrument for arguing about the trade-offs with the map in front of you.

5
criteria combined into a single weighted suitability score
70%
of the score set by terrain alone — slope and aspect
4
suitability classes, from ‘very good’ to ‘avoid’
5,600 km²
of the West Bank scored, cell by cell
Next project
Three summers of exploration fieldwork across the Canadian North →