Skip to content
Ed Lewis edited this page Nov 28, 2019 · 21 revisions

PACT: Protected Area Coverage Tool.

This document is to guide you through the usage of PACT, an analytical tool developed to create geospatial products and statistics on the World Database of Protected Areas.


Author: Ed Lewis
Supervisor: Yichuan Shi


Index

(Insert Index)


Background

The WDPA team have been running analyses on the WDPA for years. Until 2014 these were infrequent, entirely manual and extremely resource intensive. Between 2014 and 2016 the analyses were automated using Esri Modelbuilder, evolving to be run in ArcGIS Pro and at a monthly interval. From 2019 it was decided to transition away from Modelbuilder into a purely python environment. There were three major reasons for this: 1) the workflow needed to be automated to a degree beyond Modelbuilder's capabilties, 2) the model was so big that it was becoming sluggish to edit, even in ArcPro and 3) the workflow and model itself needed be more publicly available, documented and editable online. It is hoped that from early 2020 all monthly analytical components of the WDPA are run via python scripts, stored in this public, centralised and regularly updated repo.


Installation requirements

This script is written to be run in ArcGIS Pro. Due to the size of the input data it is recommended that your computer has at least 20gb RAM and >50gb space on the C drive.


A description of the tool's inputs

This tool has 10 inputs that need to be prepared and defined prior to running the script:

  1. WDPA Public point available here;
  2. WDPA Restricted point (CHN) (not available to non-UNEP-WCMC staff);
  3. WDPA Public polygon available here;
  4. WDPA Restricted polygon (CHN) (not available to non-UNEP-WCMC staff);
  5. WDPA Restricted polygon (CDDA) (not available to non-UNEP-WCMC staff);
  6. WDPA Restricted polygon (SHN) (not available to non-UNEP-WCMC staff);
  7. PAME sites available here
  8. Baselayer available here;
  9. Baselayer tabular available here;
  10. A Projection file (.prj) for outputs being projected into Mollweide.

How PACT works

**Please Note that this description covers how the model works at a high level. To know which specific tools are used, and the specific parameters of those tools then please review the script itself.

Fundamentally, the tool can be divided into 7 major steps:
Step 1: Preparing the WDPA for analysis;
Step 2: Geoprocessing the entire WDPA;
Step 3: Geoprocessing the entire WDPA on a country by country basis;
Step 4: Geoprocessing sites in the WDPA with a PAME assessment;
Step 5: Creating summary statistics on the entire WDPA;
Step 6: Creating summary statistics on the entire WDPA on a country by country basis;
Step 7: Creating summary statistics on sites in the WDPA with a PAME assessment.

There is currently some duplication amongst these steps, but as described below, there is a reason for this.
We are continually trying to streamline the workflow in this methodology.

Step 1: Prepare the WDPA

In all WDPA analyses we remove certain sites. These are sites with a STATUS of 'Proposed' or 'Not Reported', sites with DESIG_ENG of 'UNESCO Man and Biosphere Reserve' and points without a REP_AREA value above 0. For STATUS sites, the reason is evidently that they are do not certainly protect the areas they cover. UNESCO MAB sites are removed as they have a three-tiered nesting structure, only the innermost of which is considered 'protected'. The sites in the WDPA do not yet specify which tier they are. Points with no REP_AREA are automatically removed in the buffering process.

The model begins by merging all the WDPA polygon inputs together and the WDPA point inputs together. These two merged files are then geometrically repaired. The repaired feature classes (fc) are then filtered to remove the sites listed above. The filtered points are then geodesically buffered in accordance to their REP_AREA values. The buffered points are then merged to the filtered polygons to create one definitive protected area fc 'PolyBuffpnt'.

Step 2: Geoprocessing the entire WDPA

This is one of the biggest steps.

The PolyBuffpnt undergoes a Union with itself, splitting up the WDPA into all overlapping component parts. (Insert image?) The XY centroid of each polygon is calculated. Those polygons with overlapping XY centroids are removed, dissolved flat to remove their overlaps and then put back. The now flat WDPA layer is now intersected with the baselayer The intersected flat WDPA layer is projected into the equal area projection Mollweide The projected layer has the area of each polygon calculated

Step 3: Geoprocessing the entire WDPA by country

Q: Why do we geoprocess again by country? **A: When we geoprocess the entire WDPA (step 1) we disregard what country submitted what site. When we report country stats we do not define a country's network based on geographic delineations (i.e. country boundaries), instead we report based on all sites that have that country's ISO3166 code. The major reason for this is PA boundaries can often be of variable accuracy and rarely align to country boundaries (which can also be disputed), thus you can easily artificially insert false negatives/positives into country reporting this way (it's also prone to political disagreement).

This branch of the model starts with the PolyBuffpnt and has a very similar set of tools as Step 2.

Sites with a single ISO3 code (i.e. 'BRA') are selected from PolyBuffpnt into a separate fc 'non-transboundary' Site with multiple ISO3 codes (i.e. 'BRA;ECU') are selected from PolyBuffpnt into a separate fc 'transboundary' Transboundary sites are erased from the non-transboundary sites, leaving geographies originating solely from transboundary sites. Both the non-transboundary and the erased-transboundary sites are intersected with the baselayer Both the non-transboundary and the erased-transboundary sites are projected into Mollweide. Both the non-transboundary and the erased-transboundary sites have their areas calculated.

Step 4: Geoprocessing sites in the WDPA with a PAME assessment;

Q: Why do we geoprocess again for PAME sites?
**A: Despite Step 4 consisting of an identical workflow to Step 3 it must be replicated with just those sites that have had a PAME assessment because...

This step also starts at the PolyBuffpnt.

The Polybuffpnt is joined to a list of sites that have had PAME assessments PAME Sites with a single ISO3 code (i.e. 'BRA') are selected from PolyBuffpnt into a separate fc 'PAME non-transboundary' PAME Site with multiple ISO3 codes (i.e. 'BRA;ECU') are selected from PolyBuffpnt into a separate fc 'PAME transboundary' PAME Transboundary sites are erased from the PAME non-transboundary sites, leaving geographies originating solely from PAME transboundary sites. Both the PAME non-transboundary and the PAME erased-transboundary sites are intersected with the baselayer Both the PAME non-transboundary and the PAME erased-transboundary sites are projected into Mollweide. Both the PAME non-transboundary and the PAME erased-transboundary sites have their areas calculated.

Step 5: Creating summary statistics on the entire WDPA

This step starts from the end of Step 2. Global reporting is quite simple, we just need to know the total global flat area of protected areas (km2), split by land, EEZ and ABNJ.

Global summary statistics are run by summing the PA area field () by realm (type field). This output file has to be further formatted however in the global summary pa coverage template - which has ~7 further formatting steps.

**Q: Why do the global coverage numbers have to be further formatted?
**A: Formatting the data in Excel enables the user to eyeball the data, compare it to the previous months' data, and ensure it is in the correct format for Informatics.

Step 6: Creating summary statistics on the entire WDPA on a country by country basis

This Step starts from the end of Step 3.

National summary statistics are run by summing the PA area field () by ISO3 and realm (type field). This output file has to be further formatted however int eh national summary pa coverage template - which has ~n further formatting steps.

**Q: Why do the national coverage stats have to be further formatted?
**A: Formatting the data in Excel enables the user to eyeball the data, compare it to the previous months' data, and ensure it is in the correct format for Informatics.

Step 7: Creating summary statistics on sites in the WDPA with a PAME assessment.

This Step starts from the end of Step 4.

The summary statistics on sites with a PAME assessment is actually embedded within Step 6 and the output there.

Step 8:


FAQs/Troubleshooting

Q: Why do UNEP-WCMC country stats differ from official nationally reported stats?
**A: There are many many reasons why there may (or more likely will) be differences between the statistics PACT creates and those produced by the country. From experience of dealing with multiple country queries on this matter, in a rough descending order of importance: which sites are included (we remove MAB/proposed/not reported whilst they will often not have regional/international), whether they include OECMs, how they define marine sites, what basemap they use, what projection they use, how they deal with points etc etc.

Q: Why don't we use geographic delineation for ABNJ coverage?
**A: ABNJ coverage is calculated through the WDPA ISO3 field because whilst many nationally designated marine protected areas may overlap (to various degrees) with the geographic ABNJ area they are in fact still managed by a country and are subject to that country's legislation. If we were to use the geographic ABNJ then the coverage of international waters would be higher, but marine areas within national jurisdiction would be lower.

Q: Why do we remove Antarctica (ATA) from the terrestrial global coverage statistics?
**A: UNEP-WCMC has been removing ATA land from global coverage statistics since the first protected planet report. Though the rationale is inconclusive, it would seemingly be due to the fact that ATA land is already largely considered some form of (reserve) [https://en.wikipedia.org/wiki/Antarctic_Treaty_System]. If ATA land were to be added then terrestrial coverage would drop quite significantly.


Exploring the outputs

There are only a few tangible outputs from PACT for two major reasons:

  1. a lot of the outputs are stored in_memory\ and are thus removed once the script has successfully completed.
  2. If not for in_memory\ the tool will create a sizeable amount (>10gb) of spatial data each time it is run, which can quickly use up space.

There are several spatial outputs that are kept as they are useful for any additional analyses.

  1. "all_wdpa_polybuffpnt": The singular spatial representation of all protected areas for that month (excluding sites for QC reasons). This is the best fc to use for doing analyses like 'counts' of particular designations etc.
  2. "all_wdpa_polybuffpnt_union_flat_intersect": The global flat version of 'all_wdpa_polybuffpnt' which has been intersected with a basemap. This is the best fc to use for doing geospatial analyses on area where you don't care about attribute data.

There are two major tabular outputs:

  1. "global_coverage_statistics": A summary table of global PA coverage (km2) per land, EEZ and ABNJ.
  2. "national_coverage_statistics": A summary table of national PA and PAME coverage (km2) by country and terrestrial/marine.

Please note that this is a workflow that is run often and is subject to change at any point. For further information please contact [email protected].