For assignment 3, you are to work in teams of four or five people.
You are asked to design a scientific experiment that is relevant to tackling climate change. You won’t be expected to perform the experiment, but you will be asked to think about what (new) information the experiment would provide, and what weaknesses and limitations there would be on interpreting the results of the experiment.
Remember, no experiment is perfect – each experiment is a compromise in which we have to balance our goals (what information would we really like to get) against what is feasible (what experiments can we actually perform using the tools we have available today). Finding a good compromise is the tricky part.
Note: your experiments will require computer models, because direct experimentation on the world itself is hard when it comes to issues that are as large as climate change. In most cases, any process that’s small enough to study in a laboratory experiment is probably too small to matter. Most of the interesting questions focus on what happens when we do things at global (or at least regional) scales.
- Choose a scientific question that you regard as important for assessing some aspect of climate change. Your chosen question should be relevant to the decisions we have to face in the coming years in dealing with climate change. For example, the question could be about what kinds of impact climate change is likely to have, and how much uncertainty there is about them; or it might be about the effectiveness of different kinds of response to climate change.
- Identify which processes in the world are the most important for answering your chosen question. For example, does the question depend on:
- the detailed physics of the greenhouse effect (e.g. how greenhouse gases affect the earth’s radiative balance)
- how changing temperatures affect other natural systems (such as weather, oceans, or ecosystems)
- the details of how human systems make use of energy (e.g. for manufacturing, agriculture, transport, housing)
- how international agreements work (e.g. setting, measuring or adjusting targets)?
- detailed economic mechanisms (e.g. energy prices, markets, taxes, incentives)?
- human behaviour (how people make decisions, how people share information, etc)?
- the properties of different technologies (e.g. energy systems, transport, computing, biotech, waste disposal, etc)?
- Identify what kinds of computer models are available that cover the processes you identified as important in step 2. What are the limitations of these models? Which of the processes that matter for your overall question are represented well, and which are represented poorly (or not at all)? Weighing up these limitations, decide which model (or models) you should work with. Note: if you get stuck at this point, you may need to go back to step 1. Not all questions can be answered using the models available today.
- Sketch out a design for an experiment with the selected model that would help answer (some part of) your overall question. Consider:
- What are your initial hypotheses for the experiment?
- What run(s) will you need to do with the model to test your hypotheses?
- What data will be needed for the model to work?
- What kind of data will you need the model to output?
- How would you analyze this data – for example what graphs might you draw to understand the results?
- Write up a report describing what you did in steps 1-4. Your report should include:
- How well you think your experiment addresses your chosen question;
- What compromises you had to make in designing the experiment;
- Which model you chose to use and why;
- The details of your experiment: your hypothesis, the run(s) that you would need, the data you would use, and how you would present the results;
- Any limitations on interpreting your results, for example how accurate are your results likely to be, and what kinds of uncertainties will there be?
- How your ideas about a suitable question evolved in the process of finding a suitable model and deciding what experiment you could do;
- Any lessons you can draw about the relationship between the questions that matter to society and the kinds of experiments we can currently perform with computer models.