testing functions in sf package

April 2020

This is related to an overlap analysis between KBAs and Protected Areas (PAs) - see the repo for current code.

Go to fragments folder to see the code and results for the below.

The aim is to calculate the increase in coverage of each KBA by PAs (as 'proportion of KBA area protected') over time. Most PAs have a year of designation in the dataset -> we can use this as the year that the part of the KBA covered by that PA was protected.

Sometimes multiple PAs overlap within a KBA -> we use the earliest year as the year that this part of the KBA was protected.

However, some PAs have no year of designation - it's missing in the dataset. This is represented by year = 0. To still be able to include them in the analysis, we assign these PAs a random year (and re-run the analysis e.g. 1000 times).

It's very computationally intensive to assign the random year and then run the overlap analysis, so we want to improve the code to:

• make a table of 'fragments' - areas of the KBA where PAs overlap it
• record the area of each fragment, and the PAs overlapping it - along with their year of designation (including where year = 0 as this will be randomised)

This way, we only have to run the spatial KBA-PA overlap once to get the fragments and their areas. Then every time we randomise the years of any PAs with year = 0, we just do a simple tabular analysis to calculate the building up of PA coverage for each KBA over the years.

• Split the KBA-PA overlaps into fragments, and for each fragment, record:
  • KBA id
  • PA id(s)
  • year(s) of PA designation - including 0
  • fragment area as proportion of the KBA

**1 - ** Make polygons from coordinates, convert them to sf objects.

• Pure polygons (sfg) -> geometry objects (sfc) -> full sf objects including data frame (sf)
• **2 - ** Crop PAs to area within the KBA
• **3 - ** Chop the PAs into fragments - where each fragment is a unique area of overlap between different PAs (like the sections of a Venn diagram where the circles represent the PAs)
• **4 - ** Make data frame A with 1 row per fragment, including the origins of the fragment (= which PAs combined to make that fragment)
• **5 - ** Make data frame B with multiple rows per fragment, recording the various PAs and their years.
• 

Then during the randomisation, we can:

• Assign a random year to any PAs with year 0
• Fill this year in in data frame B
• For each fragment, get earliest year of protection (may be different when it involves a randomised PA)
• Order fragments by year
• Calculate build up of the proportion of the KBA protected in each year.