search

pacificclimate / climate-explorer-frontend

6 stars 1 forks source link

Reconsider matching graph timeseries resolutions on extreme precipitation portal #144

Open corviday opened 6 years ago

corviday commented 6 years ago

At present, Climate Explorer, when comparing two variables on the same graph, matches the resolution of the secondary variable to the primary variable. My thinking on this was that comparing variables would only make sense if they had the same time resolution.

However, I'm not sure this completely holds for the Extreme Precipitation portal. Many of the climdex values are annual-only, representing the number of days in a given year that whatever precipitation related thing happens. Upon playing with the annual climdices in the Extreme Precipitation portal, I think it would actually be more helpful to compare values of the form [days in a year x precipitation thing happens] against a timeseries showing monthly or seasonal precipitation (or even daily, but we don't have that) than against a single annual total precipitation value.

Perhaps the difference is that, in terms of representation inside a datafile, the annual-only climdexes have one value per year, but that single value actually encapsulates a number of higher resolution ones (binary classifications of each day), so it makes sense to compare them with a higher-resolution secondary variable timeseries.

Plus, c3 just draws both annual timeseries atop each other and it looks like an error. linesoverlapping

corviday commented 6 years ago

Possibly add a prop toDualDataController that would control whether or not it limited secondary variable resolutions to resolutions available in the primary.

I still feel like matching resolutions is probably the right way to handle comparing non-derived variables like temperature and precipitation.

corviday commented 6 years ago

Summary: The underlying question here is: when is it useful to compare two variables measured at different resolutions on a graph?

At present, climate explorer treats the answer as "never", resulting in these flatline graphs where annual-resolution climdexes are only be compared with annual-resolution precipitation values, even when higher-resolution precipitation data is available.

jameshiebert commented 4 years ago

@tqmurdock Do you have any opinions on this issue?