From Spatial Data to Decisions: Building a Suitability Model with ArcGIS Pro
A practical exploration of Multi-Criteria Suitability Analysis through the Esri MOOC: Spatial Data Science — The New Frontier in Analytics
1. Introduction: Turning a Geographic Problem into a Spatial Decision
One of the most interesting parts of my recent learning experience with Esri was working on suitability modeling.
I completed this practical exercise as part of Section 3 of the six-week MOOC Spatial Data Science: The New Frontier in Analytics. The objective was to identify suitable locations for a carbon-neutral corporate headquarters in Vermont, USA.
At first, the problem seemed straightforward: find a location with enough available land and convenient access to infrastructure.
However, choosing a suitable location involves much more than identifying empty spaces on a map.
A potential site must satisfy several conditions simultaneously. It should have appropriate terrain, suitable land use, reasonable accessibility, access to amenities, opportunities for renewable energy, and minimal environmental impact.
This is where Multi-Criteria Suitability Analysis becomes useful.
Instead of evaluating each geographic factor independently, suitability modeling provides a framework for integrating multiple spatial datasets, assigning preferences, evaluating their relative importance, and identifying locations that best satisfy a defined objective.
In this article, I will walk through the workflow I followed in ArcGIS Pro, explain the analytical decisions behind each stage, and discuss what I learned from the final results.
2. Understanding the Problem and Defining the Criteria
Before opening the analytical tools, the first important step was understanding what we actually wanted to achieve.
The exercise organized the problem into four main objectives:
1. Building Site
Identify land that is physically appropriate for construction by considering slope, land use, accessibility, and visibility.
2. Energy Independence
Identify locations with favorable conditions for renewable energy production, particularly solar radiation potential, proximity to electrical infrastructure, and terrain orientation.
3. Access to Amenities
Consider proximity to important services and facilities, including the airport, recreational areas, and residential areas.
4. Environmental Concerns
Avoid locations that could negatively affect protected areas, biological areas, wetlands, and streams.
What I found particularly interesting here was the way a general planning problem was translated into measurable geographic criteria.
For example, instead of simply saying that a site should have good accessibility, we can calculate its distance from major roads. Instead of saying that construction should avoid steep terrain, we can derive a slope dataset from a digital elevation model.
This translation from real-world requirements into spatial criteria is the foundation of the entire analysis.
3. Deriving Spatial Criteria from Existing Data
Once the criteria were defined, the next step was preparing the datasets required for the analysis.
This stage is important because the quality of a suitability model depends heavily on the quality and relevance of its input data.
3.1 Deriving slope from a Digital Elevation Model
One of the first operations I performed was deriving slope from the elevation dataset.
A Digital Elevation Model (DEM) represents the elevation of the terrain. However, elevation values alone do not directly tell us whether an area is appropriate for construction.
Slope provides additional information by measuring how steep the terrain is.
In ArcGIS Pro, I used the Surface Parameters tool with the following configuration:
- Input surface raster: Elevation_30
- Output raster: Slope
- Input analysis mask: Study_Mask
- Parameter type: Slope
- Local surface type: Quadratic
- Neighborhood distance: 30 meters
- Slope measurement: Degree
The resulting raster represents the spatial variation of terrain steepness across the study area.
From a construction perspective, flatter areas are generally more favorable because steep terrain can introduce additional engineering difficulties and construction costs.
This is a good example of how an existing geographic dataset can be transformed into a new analytical criterion.
3.2 Calculating Distance from Major Roads
Another important criterion was accessibility.
For this purpose, I used the Distance Accumulation tool to calculate distance from the major road network.
However, there was an additional consideration: geographical barriers.
The Green Mountain Ridge was introduced as a barrier so that the distance analysis could account for geographical obstacles rather than treating every direction as equally accessible.
The resulting distance raster, Dist_MRoads, represents the accumulated distance from major roads while considering the specified barrier.
This distinction matters because straight-line proximity does not always represent realistic accessibility.
Two locations may appear close on a map but have very different accessibility conditions because of the terrain or other geographical constraints.
3.3 Other Derived Criteria
The exercise also incorporated visibility analysis from major roads, represented by the Views_Zero dataset.
Together, these derived layers provided additional information for evaluating the building-site objective.
At this stage, the main idea became clear to me: spatial analysis is not simply about collecting existing layers. Sometimes, we need to generate new information from the available data before we can use it in a decision-making model.
4. Transforming Criteria into a Common Suitability Scale
This was one of the most important parts of the exercise.
Different geographic datasets do not necessarily have comparable values.
For example:
- Slope is measured in degrees.
- Distance is measured in meters.
- Land use contains categorical values.
- Visibility represents another type of spatial information.
It would not be meaningful to combine these raw values directly.
To solve this problem, the Suitability Modeler in ArcGIS Pro allows criteria to be transformed into a common suitability scale.
In this exercise, the selected scale was 1 to 10:
- 1 represents the least preferred condition.
- 10 represents the most preferred condition.
This transformation allows different datasets to become comparable within the same analytical framework.
4.1 Transforming Categorical Data
Land-use data is a good example of categorical information.
Different categories represent different types of land cover, such as forests, agricultural areas, developed areas, wetlands, and open water.
Each category was assigned a suitability score according to its relevance to the construction objective.
For example, wetlands and certain developed areas received lower suitability scores, while more favorable land-use categories received higher scores.
The important point is that categories do not have an inherent numerical preference. Their scores must reflect the objectives of the analysis.
4.2 Transforming Continuous Data
Continuous criteria require a different approach.
For slope, smaller values are generally preferred because flatter terrain is more appropriate for construction.
I used the MSSmall transformation function to represent this preference.
For distance from major roads, the Small function was used to represent the preference for locations closer to roads.
The visibility criterion was transformed using a LogisticDecay function to represent the intended preference for visibility conditions.
Get Mohammed Belfellah’s stories in your inbox
Join Medium for free to get updates from this writer.
These functions are important because they define how the original values are converted into suitability scores.
A transformation is not simply changing the numerical range. It is expressing the relationship between a geographic variable and the objective being evaluated.
5. Weighting and Combining the Criteria
After transforming the criteria, the next step was determining their relative importance.
Not every factor contributes equally to a planning decision.
For example, terrain conditions may be more important for construction feasibility than proximity to a particular amenity.
The Suitability Modeler provides a weighting mechanism that allows the analyst to express these differences.
In the building-site submodel, the criteria were combined using their assigned weights.
The model used a multiplier-based weighting approach.
Conceptually, the weighted suitability score can be represented as:
Weighted Suitability = Σ (Criterion Suitability × Criterion Weight)
This makes it possible to integrate several transformed criteria into a single suitability surface.
5.1 Creating Separate Objective Submodels
Rather than combining every criterion into one large model immediately, the exercise used four separate submodels:
- HQ_Building_Site
- HQ_Access_to_Amenities
- HQ_Energy_Independence
- HQ_Environmental_Concerns
Each submodel represents one of the original planning objectives.
I found this approach particularly useful because it keeps the analytical structure organized.
It also makes the model easier to understand, evaluate, and modify.
5.2 Combining the Submodels
After preparing the four objective submodels, I created a combined model named HQ_Combined_Model.
The four submodels were added as inputs,
each with a multiplier weight of 1
The final output was named HQ_Final_Suitability.
This produced an integrated suitability raster that represents the combined influence of the four objectives.
6. The Final Suitability Map
After combining the four objectives, the model generated the final suitability raster.
This is where the individual analytical stages came together.
Instead of looking at slope, land use, accessibility, energy potential, and environmental concerns separately, we now have a single spatial representation of their combined influence.
The resulting map allows us to examine how suitability varies throughout the study area.
However, producing a suitability raster is not necessarily the end of the analysis.
A map can show where favorable conditions exist, but a planning problem often requires identifying specific locations that satisfy additional spatial requirements.
This leads to the next stage: locating suitable regions.
7. Locating the Most Suitable Regions
One of the most interesting operations in the exercise was using the Locate functionality of the Suitability Modeler.
The objective was no longer simply to visualize suitability. It was to identify specific candidate regions based on defined spatial requirements.
The Locate operation used the following parameters:
ParameterConfigurationInput suitability mapHQ_Final_SuitabilityTotal area500 hectaresNumber of regions5Region shapeCircleShape/Utility tradeoff50%Evaluation methodHighest average valueMinimum distance between regions0.5 kmMaximum distance between regions2 km
These parameters introduced an additional spatial dimension to the analysis.
The model had to identify five regions while considering both suitability and the specified spatial constraints.
The shape/utility tradeoff also allowed a balance between the geometry of the regions and their suitability.
This is an important distinction: the highest-scoring individual pixels do not automatically form the most useful candidate sites.
The spatial configuration of the selected regions matters as well.
8. Evaluating the Results
After locating the candidate regions, the final step was evaluating whether the results satisfied the original objectives.
This is an important part of spatial decision-making.
A model should not be considered successful simply because it produces a visually attractive map.
The selected locations need to be examined against the original criteria and spatial requirements.
For example:
- Do the candidate regions have favorable terrain conditions?
- Are they reasonably accessible?
- Do they avoid environmentally sensitive areas?
- Do they support the energy-independence objective?
- Do they satisfy the required spatial configuration?
These questions help determine whether the analytical results are consistent with the original problem.
It is also important to recognize that suitability modeling depends on assumptions, input data, and assigned weights.
Changing the criteria or their relative importance can produce different results.
Therefore, the final suitability map should be understood as the outcome of a defined analytical scenario, not as an absolute statement that a particular location is objectively the best.
For an actual development project, additional investigation, updated datasets, field verification, regulatory checks, and stakeholder consultation would still be necessary.
9. What I Learned from This Exercise
For me, the most valuable part of this exercise was understanding how different spatial datasets can be transformed into a structured decision-support model.
Before working through this workflow, it was easy to think of GIS analysis as a collection of independent operations: calculating distances, generating slope, classifying land use, and producing maps.
This exercise showed me how those operations can be connected to solve a much larger problem.
I learned that:
Spatial criteria must come from clearly defined objectives. The analytical process should begin with understanding the real-world problem, not simply selecting tools.
Data transformation is fundamental. Different geographic variables need to be standardized before they can be meaningfully combined.
Weighting introduces analytical decisions. The importance assigned to each criterion directly influences the final suitability result.
Spatial configuration matters. Identifying suitable regions requires more than finding the highest suitability values; area, shape, and distance constraints can also influence the result.
Evaluation is essential. The final map needs to be examined in relation to the original objectives and assumptions.
Perhaps my biggest takeaway is that suitability modeling is not simply about finding the greenest areas on a map.
It is about translating real-world requirements into spatial criteria, understanding the relationships between different geographic factors, and combining them in a transparent analytical framework.
10. Final Thoughts
This exercise gave me a practical understanding of how ArcGIS Pro can be used for more advanced spatial analysis and decision support.
What I particularly appreciated was the complete workflow: starting with a real-world problem, preparing and deriving spatial criteria, transforming datasets, applying weights, combining multiple objectives, and finally identifying candidate locations.
It also made me more aware of the responsibility involved in spatial modeling. Every criterion, transformation, and weight represents an analytical choice that can influence the final result.
I see this as one of the most useful lessons from the Esri MOOC Spatial Data Science: The New Frontier in Analytics.
For anyone working in GIS, Remote Sensing, spatial planning, or geospatial data science, understanding how to move from geographic data to evidence-based decisions is an important skill to develop.
A map shows us what exists. A well-designed spatial model helps us understand where, why, and under which conditions a decision makes sense.
This article reflects my learning experience and practical work during Section 3: Suitability Modeling, as part of the Esri MOOC Spatial Data Science: The New Frontier in Analytics.
