Calculating anomalies in time series

[rdstern]

I have been stuck in preparing the video on comparing station and satellite data. The one where we produce summaries. The current text refers to the CM SAF methods. They produce simple summaries. We produce some of them, but not all.

I hope @dannyparsons can also confirm, but they work more with anomalies than we do. And so do lots of people who process time series. We tend to split up the data to look separately at each month, or day. The seemingly standard methods of analysis calculate anomalies that take out the seasonal component.

Often they only do that, and I like what we do. But we can't yet do their types of anamolies. I think we should - and quite urgently. I think it is needed for the full set of comparisons. I suggest it could also simplify our quality control methods to be able to use the anomalies, rather than the data for ce=hecking some elements.

What we need might also be quite easy, because there is an anomalize package (of course!). It fits with tidyverse and produces some ggplots.

I suggest we should evaluate and add it quickly even if it is initially not in a dialogue.

I also suggest that what we need could perhaps simply be extra buttons in the Climatic > Prepare > Transform dialogue. It is a transformation after all! We might include it too in the Climatic > Compare > Calculate dialogue, so it is an obvious option for the comparisons later.

Then we should check that this does give a simple route to the summaries used by CM-SAF.

I suggest its implementation can also be related to our need to include items from WMO 100 on climatological practices. I assumed we would need to have a time series "module", but this provision is a key elements that could be interpreted as satisfying at least part of the WMO 100 on time series.

I hope that @shadrackkibet and @Cal-Insta might be interested in this discussion as well as @dannyparsons.