The problem, the data, and consultation
Choosing a problem GIS can actually answer
- A spatial problem is one whose answer is a place, or a set of places. If the answer is a number or an opinion, GIS is the wrong tool.
- The best problems are of the form: where should X go, which places meet all these conditions, how many are inside this zone, or what is the best route between these points.
- The problem may be hypothetical at Level 3 — a proposal that has not been made, a scenario a council could face — provided the spatial data you use is real.
- Weak problems: anything with one obvious answer, anything with no criteria to combine, and anything you cannot get layers for.
- The worked problem on these pages: which land at the invented Ōtaraki is suitable for relocating the 14 properties inside the 20-year coastal erosion hazard zone? — hypothetical, spatial, and requiring several conditions to be combined.
Writing the problem as criteria
- Turn the problem into a list of conditions before you open any software. This is the step that decides whether the analysis works.
- In the invented Ōtaraki problem:
- outside the 50-year erosion hazard zone
- outside a 50 m buffer on the stream (a planning restriction)
- within 1 km of the existing road and service network
- not in the flood-prone gully
- at least 0.4 ha per site
- Each condition becomes a layer or a query, which is why writing them first makes the whole analysis obvious.
- Say where each threshold came from. 50 m because that is the district plan's riparian setback; 1 km because beyond that new services cost more than the land is worth.
- A criterion with no source is an assumption, and an assumption you have to defend later.
Getting real spatial data
- Real data is required, even where the problem is hypothetical.
- Where to look: council and regional council open-data portals (hazard zones, zoning, property boundaries), LINZ (topographic, parcels, aerial imagery), Stats NZ (statistical areas), and your school's GIS licence.
- Record the source, the date and the coordinate system for every layer. You will need all three in your evaluation, and mismatched coordinate systems are the commonest reason a GIS analysis silently produces nonsense.
- Check the resolution matches the question. A hazard layer mapped at 1:50,000 cannot answer a question about individual property boundaries, and using it as if it could is the error that matters most in this standard.
- You may collect your own spatial data too — GPS points from fieldwork are real spatial data and integrate well with downloaded layers.
Consultation at Level 3
- You initiate the discussion, and it is specifically about which geospatial technique to use.
- Bring your teacher three things: the problem written as criteria, the layers you have found, and a proposed technique with a reason.
- Good consultation questions: Is a 50 m buffer the right tool here, or should I be using a distance surface? Should I intersect all five layers at once or in stages? Is my hazard layer's resolution fine enough for a property-level answer?
- Record what you asked, what was suggested and what you decided. That record is evidence of the process and it feeds directly into the evaluation.
- Consultation is not permission. You can decline advice provided you can say why — and doing so, with a reason, is stronger evidence of understanding than following it.
Setting up the project properly
- Put every layer in the same coordinate system before you do anything else. New Zealand data is usually NZTM2000; mixing it with WGS84 without transforming produces results that look plausible and are wrong.
- Name your layers so they mean something. hazard_20yr, stream_buffer_50m, parcels_2024 — not layer1, layer2.
- Keep the original data untouched and work on copies, so a mistake is recoverable.
- Write down each step as you do it. You must explain and then evaluate the manipulations, and reconstructing them from memory afterwards is where reports lose detail.
- Take a screenshot at each stage. They are the evidence that you did the manipulation and they make the report much easier to write.
Worked Example
Worked example
A student proposes: "I'm going to use GIS to map the erosion at Ōtaraki Beach."
Diagnose the problem, then rewrite it and set up the analysis.
Answer:
Step 1 — test whether it is a problem at all.
Mapping the erosion is a presentation task, not a problem. It has no criteria to combine, no decision to inform, and its answer is a picture rather than a place. GIS can do it, and doing it would not meet a standard about solving a geographic problem.
Step 2 — find the decision hiding behind it.
The erosion matters because 14 properties are inside the 20-year hazard zone and may have to move. The real spatial problem is: where could they go? That has a spatial answer, requires several conditions to be combined, and is exactly the kind of question GIS exists for.
Step 3 — write the problem as criteria, with a source for each threshold.
Find all land at Ōtaraki that is:
- outside the 50-year erosion hazard zone — source: regional council hazard layer, 2023
- outside a 50 m stream buffer — source: district plan riparian setback
- within 1 km of the existing road and service network — source: cost threshold from the council's infrastructure report
- not within the flood-prone gully — source: council flood layer
- at least 0.4 ha per site — source: the minimum lot size in the relevant zone
Step 4 — list the layers each criterion needs.
Hazard zones (regional council); stream centrelines (LINZ); roads and service lines (council); flood extent (council); property parcels with area attributes (LINZ). Five criteria, five layers — and the match between them is the sign the criteria were written properly.
Step 5 — check the two things that silently break a GIS project.
Coordinate systems: all five layers reprojected to NZTM2000 before any processing. Resolution: the hazard layer is mapped at a scale fine enough for property-level use — if it were not, the honest answer would be to report a zone of suitable land rather than a list of individual parcels.
Step 6 — the consultation questions to take to the teacher.
Is a fixed 50 m buffer right, or should the setback vary with slope? Should I intersect all five layers at once, or apply the hard constraints first and the preferences second? Is my hazard layer fine enough to name individual parcels, or should I report an area?