Gains:
- Ability to define the GIS logic of land use, suitability and accessibility analysis and the role of AI in this workflow
- Ability to use the concepts of layer, attribute and coordinate system correctly when designing analysis fiction with AI
- Ability to verify AI suggested analysis steps with real data quality, spatial accuracy and coordinate reference
The engine of city planning is spatial data: what's where, how much there is, how far apart. The software family that collects, stores, queries and maps this data is called Geographic Information Systems (GIS for short; GIS in English). A land use analysis answers questions such as where a school will be built, whether an area is suitable for housing or industry, how many people a green area serves, etc. with GIS. AI is a powerful assistant in designing the setup of these analyses, sequencing the steps, arranging the criteria and weights, and reporting the result. But the analysis is based on spatial data, and the quality, coordinate system and currency of the spatial data determine the result. AI cannot see this data; It only edits the fiction you provide. Therefore, every step suggested by AI must be tested with real data and accurate coordinate reference.
Three basic concepts of GIS
Layer: Each data type is a separate layer — roads are one layer, buildings are another, slope is another. Analysis is overlaying and relating layers.
Attribute: Every spatial object has a table — the area, use, owner status of a parcel. The spatial question is often an attribute question: "parcels whose use is 'green space' and whose area is greater than 5000 m²".
Coordinate reference system (CRS): It is the mathematical framework that turns the curved surface of the Earth into a flat map. ITRF/TM (e.g. national systems defined by EPSG codes) are widely used in Türkiye. False CRS is the most common and most insidious spatial error: layers do not overlap, areas and distances are completely wrong. If AI suggests distance or area calculation, the first check is always CRS.
Caution: Do not assume that a distance or area result is correct just because it seems reasonable. With incorrect CRS, two layers may appear to "overlap" and slide by meters. Always compare areas in m²/hectare with a known reference (e.g. the title area of a parcel).
The logic of land use and suitability analysis
A suitability analysis scores how suitable each location is for a purpose (for example, new housing). Typical steps:
- Set the criteria. Slope, flood risk, proximity to transportation, current use, protection areas.
- Score each criterion. For example, slope 0-5% = 10 points, 5-15% = 6, 15%+ = 1.
- Lose weight. If the flood criterion is very important in the residence, its weight will be high.
- Merge layers. Assign a fitness score to each cell/field with a weighted sum.
- Comment and verify. Test the score with field knowledge and legislation.
AI is very useful in discussing this criterion-weight construct, reminding you of a forgotten criterion, and organizing the scoring table. But remember that weights are value judgments: whether to give more weight to flooding or transportation is a public choice, not an AI decision.
Step by step: Constructing a suitability analysis with AI
- Define the purpose and domain. "New housing site availability in district X."
- Take a data inventory. List the layers you have, their sources, dates, and CRS.
- Have AI draft criteria-weights. Ask if there are any missing criteria.
- Create the scoring table together. Clarify thresholds for each criterion.
- Translated the workflow into GIS steps. Which layer, which process, in what order.
- Check data quality. Timeliness, accuracy, CRS compliance.
- Verify the result. Field, legislation, known reference.
Weak prompt / Strong prompt
Weak prompt: "Find the best place for housing."
AI knows neither the domain, nor the data, nor the criteria; It returns a generic and useless text.
Strong prompt: "I'm setting up a suitability analysis: target new residential area, work area [district]. The layers I have: slope (from DEM), flood risk (SHW), road network, existing land use, protected areas. (a) remind me of criteria that might be missing, (b) suggest a table of reasonable scoring thresholds for each criterion, (c) mark where the weights contain value judgments. Finalize the weights for me; present options."
The second prompt makes the AI a discussion partner, keeping the decision maker with the planner.
Four copyable templates
Task: Propose a set of suitability analysis criteria for the following list of objectives and tiers. Output: criterion | why is it important | possible scoring threshold | "does it contain value judgments". Also list any criteria that may be missing. Purpose: [...] | Workspace: [...] | Layers: [...]
Task: Examine the criteria-weighting table below. (a) Check whether the sum of the weights is significant, (b) flag conflicting or duplicate criteria, (c) list assumptions that may mislead the result. Decision making; just check.Table: [...]
Task: Translate the following analysis steps into a GIS workflow. For each step: input layer(s), process type, output and coordinate system / unit point to be CONSIDERED. Writing code; just describe your workflow.Steps: [...]
Task: Translate the following data inventory into a quality control chart.Columns: layer | source | date/current | CRS | known limit/risk of error.Mark layers that need to be queried before use.Inventory: [...]
A table: analysis type and validation
Analysis
The question he asked
The role of AI
critical verification
Suitability
Where is it suitable?
Criterion-weight construct
Weight = public choice; C.R.S.
accessibility
How close/accessible?
Method selection outline
Network data update
Density/coverage
How much is there, how is it distributed?
Class and threshold recommendation
Population/area data source
change
What has changed?
comparison fiction
Date and CRS compliance
buffer/distance
What's within the border?
distance logic
Unit and CRS
three mini cases
Case 1 — CRS trap. An intern overlaps two layers; the map appears to "overlap". When the AI analyzes a buffer it prints, the 500-meter buffer turns out to be strangely small. Problem: one layer is in geographic (degree) CRS, the other is in metric CRS; 500 "degrees" are meaningless. After projection to the correct CRS, the result improves. This is one of the most invisible of the most expensive mistakes.
Case 2 — Forgotten criterion. A team uses slope, flood and accessibility as criteria for housing suitability. Have the AI oversee the editing; AI "Have you added the geological/disaster risk (fault, landslide) criterion?" he reminds. The team realizes that they missed this critical criterion. The AI didn't make a decision, it closed a blind spot.
Case 3 — Weight transparency. A municipality submits a suitability map to the council. AI clearly prints in the report draft that each weight is a preference and the map will change with different weights. The council asked "what happens if we increase the flood weight?" he becomes capable of asking. Transparency transforms analysis from an imposition into a tool for discussion.
Common mistakes
- Not checking CRS. The most common and expensive mistake in spatial analysis; Check before each distance/area operation.
- Mistaking the weights for technical facts. Weight is a value judgement; It must be transparent and remain debatable.
- Ignoring data currency. The old land use layer misleads today's judgment.
- Mistaking AI for a data source. AI cannot see the layer; edits the plot. Data always comes from real, verified source.
- Not testing the result in the field. An area may appear “suitable” on the map, but on the ground it may be a completely different reality.
In summary
GIS is the spatial data engine of planning; layer, attribute and coordinate system are its three basic concepts. Suitability analysis scores each location for a purpose with criteria, scoring and weights. AI is a powerful aid in designing this setup, reminding of missing criteria, sorting the workflow and reporting; but he cannot see the data and make a decision. The most insidious mistake is the wrong coordinate system; Weights are not technical but value judgments and should be transparent. Verify every result with up-to-date data, accurate CRS and field information.
Application task
Choose a purpose (for example, a new nursery location). (1) Make an inventory of the layers you have or assume with the fourth template; Write the CRS and freshness status of each layer. (2) With the first template, suggest a set of criteria-weights to the AI and capture at least one missing criterion. (3) Mark which weights are value judgments and write in one sentence how you would explain them to the assembly. (4) Describe how you will verify the result in the field.
checklist
- [ ] I checked and synchronized the coordinate system (CRS) of all layers.
- [ ] I set the criteria and had the AI scan for missing criteria.
- [ ] I marked that the weights are value judgments and kept them transparent.
- [ ] I evaluated the data currency and accuracy.
- [ ] I used AI as an editing assistant, not as a data source.
- [ ] I verified the result with a known reference and field knowledge.
- [ ] I saved the data inventory and verification trace.