Proposing, justifying and evaluating the solution
What the three grades ask for
- Achieved: propose a solution, supported by evidence from the analysis.
- Merit: justify the solution, showing why it is better than the alternatives.
- Excellence: fully justify it, with clarity of argument and holistic understanding, and evaluate the manipulations as well.
- Note that Excellence requires both halves. A brilliant justification with unevaluated manipulations does not get there, and neither does a careful evaluation of the software work with a thin recommendation.
- The solution must follow from the analysis, not from your prior opinion. If your recommendation could have been written before the GIS work, the GIS work was decoration.
Proposing a solution
- State it as a place, or set of places, with numbers. The 11 parcels shown in Figure 4, totalling 6.8 ha, on the western terrace — not somewhere inland.
- Say how the analysis produced it: the criteria applied, the query run, and the count that resulted.
- Say what it does not cover. In the invented Ōtaraki problem, 11 parcels totalling 6.8 ha for 14 properties only works if some sites are subdivided near the minimum lot size — and that has to be said, not buried.
- Include the map. The layout must be a map, and the recommendation should be readable from it.
- Distinguish the technically qualifying set from your recommended subset, if you are narrowing further. Those are two different results and conflating them hides a judgement.
Justifying it against alternatives
- The alternatives here are other configurations of the same analysis, not other topics:
- a different threshold — a 30 m buffer, or a 2 km service distance
- a different criterion set — dropping lot size, adding slope
- a different site — the second-best cluster your analysis identified
- doing nothing — leaving the properties where they are
- Use your own outputs to test them. The elimination graph shows that relaxing the service-distance criterion would free up no additional parcels, so that alternative can be rejected on evidence rather than on judgement.
- Say which criterion is binding, because that is the one an alternative would have to change. In the invented case it is hazard extent, and relaxing that is exactly what the problem exists to avoid.
- A justification that never mentions a number from your own analysis is not a justification — it is an opinion with a map attached.
Holistic understanding
- The standard describes Excellence as clarity of argument and holistic understanding. Holistic means the solution accounts for more than the criteria you could digitise.
- Three things GIS cannot see, and all three matter here: whether the land is available (someone owns it), whether it is acceptable (people have to want to move there), and whether it is fundable.
- Name them explicitly. My analysis identifies where relocation is physically and legally possible; it does not identify where it is achievable, and those are different questions.
- Connect back to the people. In the invented case the answer is the same coast as the AS91431 issue: the 11 parcels are a technical answer to a question that is, for the households involved, about whether to leave at all.
- Acknowledging the limits of the method is not weakness — it is the difference between a result and a recommendation.
Evaluating the manipulations
- Say what each manipulation assumed. A buffer assumes the setback is uniform; an erase assumes the boundary is exact; a query assumes the criteria are independent.
- Test your weakest input rather than describing it. Re-run the analysis with the modelled hazard line offset by its plausible uncertainty and report how many results change.
- Report the manipulation that contributed nothing and say what that tells you about the site.
- Distinguish a wrong technique from a crude criterion. A fixed buffer correctly reproduces a crude planning rule — the manipulation is right and the rule is blunt, and those are separate judgements.
- Close with what you would do differently, in terms of technique rather than effort.
Worked Example
Worked example
Propose and fully justify a solution to the invented Ōtaraki relocation problem, and evaluate the manipulations used.
Answer:
Step 1 — state the solution as a place, with numbers.
The analysis identifies 11 parcels totalling 6.8 ha on the western terrace, all outside the 50-year hazard zone, outside the 50 m stream buffer and the flood extent, within 1 km of the service network, and at least 0.4 ha in area. I recommend the northern cluster of 7 of those parcels (4.4 ha) as the relocation site.
Step 2 — distinguish the qualifying set from the recommendation, and say what narrowed it.
The 11 parcels are what the query returned. The 7 I recommend are the subset forming a contiguous block, which the criteria did not test for. Contiguity matters because a relocated street needs one service extension rather than four, and because moving a community that stays together is a different proposition from dispersing it. That is a judgement I have added to the analysis, and I am flagging it rather than presenting it as an output.
Step 3 — justify against the alternatives, using the analysis's own numbers.
- Relaxing the service-distance criterion to 2 km: the elimination graph shows this criterion removed 0 parcels, so relaxing it frees up nothing. Rejected on evidence.
- Relaxing the lot-size minimum: would add parcels, but the recommendation already requires subdivision near the minimum, so this compounds a problem rather than solving one.
- Relaxing the hazard criterion: would free the most land — 9 of 16 failures were on hazard — and it is the one criterion that cannot be relaxed, because avoiding the hazard is the entire purpose of relocating.
- The second cluster (4 parcels, 2.4 ha, southern terrace): qualifies technically but is separated from the service network by the gully, so servicing it costs more than its size justifies.
- Doing nothing: leaves 14 properties inside a 20-year hazard zone, which is the situation the analysis exists to address.
Step 4 — the holistic step: what the analysis cannot see.
Availability: all 11 parcels are privately owned, and my analysis says nothing about willingness to sell. Acceptability: relocating households is only a solution if the destination is one they will accept, and 7 parcels subdivided close to the minimum is a materially different neighbourhood from the one they are leaving. Fundability: the service extension is not costed here. My analysis identifies where relocation is possible; it does not establish that it is achievable, and presenting the map as an answer to the second question would be a misuse of it.
Step 5 — evaluate the manipulations.
The query is reproducible, which is its strongest property. The attribute-table edit produced the most useful result — that hazard is binding and service distance is inert — and without it I would have had a correct map and no understanding of it. The shared limitation of every operation is that they treat a modelled hazard boundary as exact, so I re-ran the analysis with the line offset 25 m landward: 2 of the 11 parcels dropped out, and both are in the recommended cluster. That is the single most important number in my evaluation, and it means my recommendation should be stated as 5 robust parcels plus 2 requiring verification, not as 7.
Step 6 — the conclusion, stated at the confidence the analysis supports.
I recommend the northern cluster, subject to on-site verification of the two marginal parcels against the hazard model. The analysis has changed my expectation in one respect: I assumed lot size would be the binding constraint and it is not — hazard extent is, which means no adjustment to the relocation design frees up more land, and the real constraint is the coast itself.