Skip to article frontmatterSkip to article content
Site not loading correctly?

This may be due to an incorrect BASE_URL configuration. See the MyST Documentation for reference.

Turning a broad spatial idea into a feasible project question

1. Why the research question matters

A research question is the engine of your project. It guides what data you need, which method makes sense, how you evaluate outputs and how you explain results.

Without a clear question, a project can become a loose collection of tools: download data, run a model, make a map and then search for a story. With a clear question, the tools become part of a focused spatial analysis.


2. From topic to question

A topic is not yet a project. A project begins when you connect a question to data, method, output and evaluation.

LevelMeaningExample pattern
TopicBroad area of interestUrban heat, glacier change, tree crowns, flood damage
ProblemWhy the topic mattersHeat exposure varies within cities and affects planning decisions
Research questionWhat your project tries to answerWhere are high land-surface-temperature patterns most visible in a selected city?
Analytical taskWhat your workflow needs to doAcquire imagery, preprocess data, calculate or model an output, evaluate it
Expected outputWhat you produceMap, figure, metric, comparison, prediction or interpretation

3. What makes a good RQ?

A good SDS320 research question is usually:

The SDS110 project-planning material (lecture 4) used the FINER criteria: feasible, interesting, novel, ethical and relevant. For SDS320, add one practical test: can you imagine the main map, figure, metric or model output that would answer the question?


4. Use PICOT to sharpen the question

The PICOT framework can help you make a question more precise.

PICOT elementMeaning for SDS320
P — PopulationWho or what is studied? This can be a place, object type, land-cover class, image set or spatial feature.
I — Intervention / IndicatorWhat is measured, detected, modelled or introduced?
C — ComparisonWhat is compared? For example before/after, area A/area B, class A/class B or method A/method B.
O — OutcomeWhat result is measured or mapped?
T — TimeOver what period or observation date?

PICOT is not mandatory, but it helps reveal vague questions.

How does [indicator] differ between [comparison groups] for [population/features] in [area] during [time period]?

5. Useful question patterns

How does [spatial phenomenon] vary across [area/time/classes]?
Can [data source or method] identify [feature/change/pattern] in [study area]?
To what extent does [variable] relate to [spatial outcome]?
Where are [features/events/changes] most visible in [data]?
How well can [method] map or predict [spatial feature] from [input data]?

6. Weak vs improved examples

The examples below are illustrative.

Weak questionWhy it is weakImproved question
How has biodiversity changed?Too broad and not clearly spatial or measurable.How has bird species richness changed across selected Swiss alpine valleys between 1990 and 2020?
Can I use deep learning on satellite images?Starts with a method, not a spatial problem.Can semantic segmentation of Sentinel-2 imagery identify broad land-cover classes in a selected study area?
Which regions have worse flood problems?Ambiguous terms and unclear metric.How did mapped flood extent differ between selected municipalities during a defined flood event?
Has this glacier changed?Too close to yes/no.By how much did glacier area change between two selected observation years, and where was retreat most visible?
Is the model good?No task, metric or reference.How accurately does the model classify selected land-cover classes compared with reference labels in the study area?

7. Scope check

Before settling on a question, check:

If several answers are unclear, narrow the question before adding technical complexity.


8. Decision points

Ask yourself:

  1. Am I asking a spatial question or only using spatial data?

  2. Does the question require analysis?

  3. Can the question be answered with data I can access?

  4. What would count as a useful result?

  5. What would count as a limitation?


9. Mini task

Write three versions to improve your question:

Broad topic:
Draft question:
Improved SDS320 research question:

Then complete:

My project asks ...
It uses ...
It compares or measures ...
It produces ...
It can be evaluated by ...

10. Key takeaways