Skip to content

{targets} pipeline to mimic IBAT's disclosure report for entire portfolios

License

Notifications You must be signed in to change notification settings

mapme-initiative/disclosure-report

Repository files navigation

Lifecycle: experimental

Important

This pipelines requires the proximity-indicators branch of mapme.biodiversity to be installed.

disclosure-report

This repository contains code that mimics IBAT’s disclosure preparation report as a reproducible {targets} pipeline.

While we do not produce PDF reports here, we go beyond IBAT’s functionality in the sense that we calculate significance scores for entire portfolios using mapme.biodiversity.

The required inputs are:

  • data/locations.json: An GDAL-readable input file with vector geometries representing project locations with attributes available to specify the area of influence (see below).

The key part of this pipeline is that you can customize the area of influence (or buffer zones) to be used for your locations. For this, you need to change the values of the variables code_column and aoi_size you can find in the header of _targets.R. code_column identifies the column in the input data to be linked via the code column in aoi_size to an associated buffer size. Note, that codes not present in aoi_size will be encoded with the default value.

The pipeline then calculates the respective indicators within the area of influence for each location. These are:

  • proximity to protected areas
  • proximity to key biodiversity areas
  • number of threatened species
  • maximum value of STAR layers threat abatement and restoration

These are individually scored to their biodiversity significance in accordance to the IBAT workflow. A site also receives a total biodiversity significance score equal to the highest score of the individual components (none, low, medium, and high).

The codes are structured in the following way:

  • inputs.R: read inputs and calculate area of influence
  • mapme.R: calculate the required indicators
  • significance.R: calculation of significance scores

To run the pipeline, adjust _targets.R to point towards your input file and run the following from a shell:

$ dvc pull
$ Rscript -e 'targets::tar_make()'

Below you can find the visualisation of the pipeline’s targets:

  • The library is already synchronized with the lockfile.
graph LR
  style Legend fill:#FFFFFF00,stroke:#000000;
  style Graph fill:#FFFFFF00,stroke:#000000;
  subgraph Legend
    direction LR
    xf1522833a4d242c5([""Up to date""]):::uptodate --- xd03d7c7dd2ddda2b([""Stem""]):::none
  end
  subgraph Graph
    direction LR
    xa538c4711b193115(["significance"]):::uptodate --> x21a9e021e2a8eb5c(["significance_output"]):::uptodate
    x148bea66679381e8(["indicators"]):::uptodate --> xa538c4711b193115(["significance"]):::uptodate
    x148bea66679381e8(["indicators"]):::uptodate --> xd7196846cd1935c0(["indicator_output"]):::uptodate
    x9755545176a05140(["data"]):::uptodate --> x148bea66679381e8(["indicators"]):::uptodate
    x2ee4157c3daee3b7(["input"]):::uptodate --> x9755545176a05140(["data"]):::uptodate
  end
  classDef uptodate stroke:#000000,color:#ffffff,fill:#354823;
  classDef none stroke:#000000,color:#000000,fill:#94a4ac;
  linkStyle 0 stroke-width:0px;
Loading

About

{targets} pipeline to mimic IBAT's disclosure report for entire portfolios

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages