How can I improve this plot?

Hi all! I'm asking a little but of support for solving a small issue concerning the following plot:

I would like to improve its readability because for large time spans, all you can see are small spikes of non-zero data basically. Which is absolutely correct, but hard to interpret.

I tried to use 30 days or 60 days movmean (over a total time span of 15 years) to smooth the underlying data being plotted, but unfortunately the obtained result, despite the readability improvement, wasn't so great. All the other smoothing functions I tried completely messed up the plot, which must represent a zero-inflated value.

I was looking for a good solution half-way between a movmean and raw data. Something that can preserve the original magnitude and spikes, providing a visual improvement of the "trend". A sort of interpolation over high and low spikes of data. Something like this (sorry, I'm really bad at drawing):

Can you suggest me a way to obtain that?

I attached the test data I'm using to produce the plot above.

Answer 1

[Star Strider]

That looks like something the envelope function would return.

[Tommaso Belluzzo]

Yes... we are on the good path here. Absolutely.

I tried using peak envelope, it's good but there are two major problems for now: finding a generalized value for peaks separation and placing a constraint so that envelope cannot go below zero.

[Star Strider]

I am not certain that there is any way to constrain it, although it has a number of parameters (for example 'peak') that you might be able to manipulate to get an acceptable result.

The only other option I can suggest is the sgolayfilt function. It has a few more options, although I am not certain that it can be constrained to remain positive.