Add local_return_period method to Impact objects and new tutorial

[ValentinGebhart]

Changes proposed in this PR:

• New method climada.engine.impact.Impact.local_return_period that computes locally (i.e. per centroid) return periods for given threshold impact values (inputs to the method). Functionality analogous to climada.hazard.base.Hazard.local_return_period (underlying functions already exist, see PR Combining several interpolation functions using a new util function #918), but with impact values instead of intensity values.
• New tutorial doc.tutorial.climada_util_local_exceedance_values.ipynb that explains what is done in the methods Hazard.local_exceedance_intensity, Hazard.local_return_period, Impact.local_exceedance_impact, and Impact.local_return_period. First, a dummy hazard object is used to showcase the different settings of the methods. Then, the methods are applied to an exemplary Hazard and an exemplary Impact object.

PR Author Checklist

• Read the Contribution Guide
• Correct target branch selected (if unsure, select develop)
• Descriptive pull request title added
• Source branch up-to-date with target branch
• Documentation updated
• Tests updated
• Tests passing
• No new linter issues
• Changelog updated

PR Reviewer Checklist

• Read the Contribution Guide
• CLIMADA Reviewer Checklist passed
• Tests passing
• No new linter issues

[spjuhel]

Very nice work! 👍

I have some minor comments on the tutorial, mostly typos/rephrasing, and maybe some clarification in the first part with the mock example hazard.

• "This is a tutorial to present and explain certain methods avaibale in Hazard and Impact objects. These methods" -> "This tutorial presents methods available in ... to compute ..."
• Cell "Compute local exceedance intensities": "We first sort the centroid's intensities in descending order"
• -- : "100****years"
• -- : First centroid D then centroid C? Typo?
• -- : It took some time for me to understand. Maybe just briefly stating why we cumulate the frequencies, linking to exceedence.
• -- : last line "how do we want to estimate" (?)
• Cell "Option 1 (default setting)": "(here, 30 years)", same for "5 and 150" further in the sentence
• -- : I would swap the last two sentences on logarithmic scaling, to get the explanation right next to it (and remove the "however").
• Cell "option 2": the user wants to guess -> estimate
• -- : "Also centroid with events intensity will be assigned zero exceedance intensity." Not sure what you mean here.
• Cell "option 4": "Here, return periods will be assigned the exceedance intensity of the next smaller of the observed return periods." -> I found this ambiguous. Is it the closest return period that is smaller or the smaller return period that is above? Maybe rephrase?
• Cell "Compute local return periods" : Rephrase. Maybe in the form "Using you can find/plot the return period of events exceeding different intensities"
• Comparison of the described methods on a realistic ... : real example rather than realistic (being extra picky here ahah)
• "for 'test' return periods of" -> why 'test' ?
• "for a return period range from 5 to 250 years" -> either "for return periods ranging from" or "for a return period range of"
• "lie outside the range of observed values for this centroid (blue scatter points)"

[ValentinGebhart]

@spjuhel Thanks a lot for the comments, I have implemented them all. If you want, you can check what I added as a brief explanation about why we sort the intensities und add up their frequencies in this specific way (second paragraph of Section Compute local exceedance intensities). I hope this makes things clearer :)

[spjuhel]

Nice! Happy to approve this PR :)

[sarah-hlsn]

This is a great tutorial! It really takes the user by the hand and goes through the different methods step-by-step. I particularly like the inclusion of a very simple mock hazard alongside real hazard objects. I have spotted some minor typo/grammar issues. Otherwise happy to approve if all checks are working. Here a couple of propsed changes:

• "Further below, we apply the methods to more real Hazard and Impact objects." either real --> realistic or more real --> real
• "The Hazard object consists of two events and has a spatial extend" extend --> extent
• "This resulting cumulative frequency for each intensity then yields the intensitiy's " intensitiy's --> intensity's