Open trevorb1 opened 3 years ago
Hey @tniet, would we be able to get your opinion on this one please? @Sfusina and I discussed it this evening but weren't able to come to a conclusive result. Do you think it is better to group the only two nuclear generating provinces in Canada, or to only group provinces that are directly connected? Or if your not sure, @Sfusina and I can discuss it again and come to a decision between the two of us!
Since we are having to change the names of the regions anyways (issue #19), if the region breakdown is also an issue we can wrap it into that update which will happen soon.
Hi @trevorb1 @Sfusina - There's challenges with most ways for doing spatial (and temporal) aggregation. My suggestion is to:
Have scripts that can quickly shift aggregation structures (for example, have the input for a spatial aggregation script be a file that defines the region code and the province codes that are in each region - the script then aggregates the provinces from the raw data as specified, and can be easily re-run if changes are needed). This means that if you choose incorrectly the first time, you can easily shift/change things for another attempt, and you can also investigate the impact of choices on your results. Main challenge with this script is that it has to work for a bunch of regionalized input data pieces, not just one (demand, supply, installed capacity, capacity factor, etc.). You also have to modify the visualization scripts to accommodate different choices.
Find literature that supports your choices. We're not the first that have struggled with this, and we won't be the last. Find journal articles that have tested the impacts of spatial aggregation (https://www-sciencedirect-com.proxy.lib.sfu.ca/science/article/pii/S1364032121001519?dgcid=raven_sd_search_email as an example), and based on these references, develop strong justifications for why your choice doesn't fall into the pitfalls others have found/discussed (or why the questions you are asking aren't impacted by your choices).
Be prepared to defend this choice (with reference to the supporting literature) to our research group, your energy modelling peers at conferences and as journal reviewers, your examining committee for your thesis/dissertation, and ultimately, yourself as you need to be confident that you are justified in what you are doing. The model challenges we've been doing are a great part of this - having good, solid, answers to questions you know will come up (for example, why are you spatially combining two disconnected provinces?). Also be prepared to run some of your scenarios with different aggregation structures to confirm your choices.
Document the choice(s) of regionalization you make, clearly showing how your choices are supported by the literature (see 2). Sina has a draft background document (part of the EMI report?) for his choice but it could use strengthening and doesn't currently include strong literature justification.
Hope that helps.
Just to summarize where we are with this issue, we have completed item one in pull #39. We still need literature references to support our regionalization choices (items 2,3,4)
We have broken Canada into four regions. The Mid-East region includes Ontario and New Brunswick based on NERC, energy source commonalities (nuclear), and better energy balancing between E and ME. Ontario and New Brunswick are not connected which seems like it might introduce error when incorporating trade. Is this going to be an issue?
If it is, maybe we could break Canada into 5 regions? Leave Ontario as one region, Quebec as another region, and group all Atlantic provinces together?