Should people be bringing their own data that they can test?
Will this be on their own laptops or ours?
Should we request data in advance so we can prepare/do pre-processing? How long will this take
More time spent on discussion/reflection rather than waiting for data to finish processing…
Some climate models will be in very different formats; since Met Office team is coming, their focus will be more on their data and UKCP18
Are participants more interested in methodologies?
Should we invite methods/package developers? Ibicus?
Ask attendees their methods of interests
There should be an assessment component as well
Good to have people “bring their own problems” but do we have enough time to work with their data?
Another way to make it more relevant is focusing in on their local area/city
Demonstrate with their geographic area of interest
Also, what language do they use?
KCL participants more interested in bias correction
Identify variables of interest - discussion? E.g. humidity
Between now and then let’s run as many methods as possible on one city for one variable; potential linking to another variable like greenspace or another SPC variable
Glasgow & max temp
Currently Griff’s tests focus on max temp and rainfall for all 3 cities (London, Glasgow, Manchester); current calibration is 1981; validation 2010
Focus on our key thing: comparison of methods; let’s get as many methods ready as possible
Demo notebook: could be made out of the guidance; want to show the pipeline somehow (e.g. live run it with a smaller dataset or preload data for some step); something visual
Sophie could run through pre-processing pipeline conceptually and focus on interactions with R assessment notebook
We already have something in R (read/visualise data, calculate metrics/show plots), but it’s for assessing the data not running the pipeline on the data
Ask participants what they are most interested in - EMAIL MO, create survey!
16 Nov meeting notes: