Geospatial Decision Support Systems

A Geospatial Decision Support System (GDSS) combines spatial analysis, mathematical models, and policy rules to help decision-makers evaluate alternative scenarios.

1. Multi-Criteria Decision Analysis (MCDA)

MCDA is a framework used to find optimal locations by evaluating multiple criteria (e.g., land suitability, slope gradients, cost, proximity to markets, environmental regulations).

  MCDA WORKFLOW PIPELINE
  +-------------+    +---------------+    +-----------------+
  | Raster/      | -> | Standardize   | -> | Apply Weights   | ---\
  | Vector Data |    | Range [0-100] |    | (AHP / Matrix)  |    \
  +-------------+    +---------------+    +-----------------+     \  +-----------------+
                                                                   +> | Weighted Linear | -> Best Sites
  +-------------+    +---------------+                            /   | Combination    |
  | Constraint  | -> | Binary Mask   | --------------------------/    +-----------------+
  | Layer       |    | [0 or 1]      |
  +-------------+    +---------------+

Criterion Standardisation:

Raw values (e.g., slopes in degrees, distance in meters) cannot be added together directly.

They must be standardized into a uniform scale (e.g., \(0\) to \(100\) or \(0\) to \(10\)), where higher numbers indicate better suitability.
Constraint Masking:

Constraints are binary exclusions (e.g., water bodies, protected national parks).

Represented as \(1\) (suitable to build) and \(0\) (strictly excluded).
Synthesis (Weighted Linear Combination):

The final suitability score is calculated by multiplying each standardized criterion score by its assigned weight, and then multiplying the total by the constraint mask.

\[\text{Score} = \left( \sum_{i=1}^{n} w_i x_i \right) \times \prod_{j=1}^{m} c_j\]

Where \(w_i\) is criterion weight, \(x_i\) is standardized score, and \(c_j\) is the binary constraint.

2. Analytical Hierarchy Process (AHP)

AHP is a structured technique for organizing and analyzing complex decisions, developed by Thomas L. Saaty. It determines weights by using pairwise comparison matrices.

Pairwise Comparison Matrix:

Criteria are compared in pairs on a scale of \(1\) (equal importance) to \(9\) (extremely more important).

If Criterion A has an importance of \(5\) relative to Criterion B, then Criterion B has an importance of \(1/5\) (\(0.2\)) relative to Criterion A.
Consistency Check:

To ensure comparison judgements are logical, we calculate the Consistency Ratio (CR):

\[\text{CR} = \frac{\text{CI}}{\text{RI}}\]

Where:

\[\text{CI} = \frac{\lambda_{\text{max}} - n}{n - 1}\]
- \(n\) is the number of criteria.
- \(\lambda_{\text{max}}\) is the principal eigenvalue.
- \(\text{RI}\) is the Random Index (standard value based on matrix size).
- CR Threshold: The CR must be \(< 0.10\) (\(10\%\)) to be considered consistent. If CR \(\ge 0.10\), pairwise rankings must be revised.

3. Hydropower Suitability Analysis Exercise

We will model suitable runoff hydropower project locations using three parameters: drainage network density (Weight \(= 50\%\)), slope gradient (Weight \(= 30\%\)), and distance to transmission lines (Weight \(= 20\%\)). Conservation areas represent a binary constraint.

Prepare Input Rasters:
- flow_accum_class.tif (reclassified values \(1-10\), where \(10\) indicates high accumulation).
- slope_class.tif (reclassified values \(1-10\), where \(10\) represents optimal steep head gradient).
- grid_proximity_class.tif (reclassified values \(1-10\), where \(10\) is closest to the grid).
- conservation_mask.tif (\(0\) inside national parks, \(1\) everywhere else).
Enter Weighted Expression in Raster Calculator:

Input the equation:

(("flow_accum_class@1" * 0.50) + ("slope_class@1" * 0.30) + ("grid_proximity_class@1" * 0.20)) * "conservation_mask@1"

Save output as hydropower_suitability.tif.
Identify Candidate Locations:

Style the output with a singleband pseudocolor ramp.

Extract pixels with suitability scores \(> 8.0\) and convert them to vector points representing candidate intake locations.