This project performs a polygonal summary on tree cover loss and intersecting layers for a given input feature using SPARK and Geotrellis.

Currently the following analysis are implemented

• Tree Cover Loss
• Annual Update
• Annual Update minimal
• Carbon Flux Full Standard Model
• Carbon Flux Sensitivity Analysis
• Glad Alerts

A simple analysis which only looks at Tree Cover Loss, Tree Cover Density (2000 or 2010) and optionally Primary Forest. Users can select one or many tree cover thresholds. Output will be a flat file, with one row per input feature and tree cover density threshold.

This type of analysis only supports simple features as input. Best used together with the ArcPY Client.

sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --analysis treecoverloss --feature_type feature --features s3://bucket/prefix/file.tsv --output s3://bucket/prefix

A complex analysis intersecting Tree Cover Loss data with more than 40 layers. This analysis is used for the GFU. Supported input features are GADM features only. Output are Summary and Change tables for ISO, ADM1 and ADM2 areas.

sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --analysis annualupdate --feature_type gadm --tcl --features s3://bucket/prefix/file.tsv --output s3://bucket/prefix

This analysis follows the same methodology as the annual update analysis above, just with fewer intersecting layers. It is used to compute statistics for the GFW country and user dashboards.

Supported input features are

• GADM
• Geostore
• WDPA
• Simple Feature

Output are Whitelist, Summary and Change tables for input features. For GADM, there will be seperate sets for ISO, ADM1 and ADM2 areas. For GADM there will also be summary tables with one row per ISO, ADM1, ADM2 and Tree Cover Density, which is used for to prepare the Download Spreatsheets on the GFW country pages. To produce final spreadsheets you will need to add another post processing step.

sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --analysis annualupdate_minimal --feature_type gadm --tcl --features s3://bucket/prefix/file.tsv --output s3://bucket/prefix
sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --analysis annualupdate_minimal --feature_type wdpa --tcl --features s3://bucket/prefix/file.tsv --output s3://bucket/prefix
sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --analysis annualupdate_minimal --feature_type geostore --tcl --features s3://bucket/prefix/file.tsv --output s3://bucket/prefix
sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --analysis annualupdate_minimal --feature_type feature --tcl --features s3://bucket/prefix/file.tsv --output s3://bucket/prefix

Carbon Flux (full standard model) analysis is used to produce statistics for GFW climate topic dashboards. It uses the same approach as the annual update analysis, but with all stock and flux outputs from the forest carbon flux model. It also analyzes several contextual layers that are unique to the carbon flux model and does not include many contextual layers used in the Tree Cover Loss analyses. It currently only works with GADM features.

sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --analysis carbonflux --feature_type gadm --tcl --features s3://bucket/prefix/file.tsv --output s3://bucket/prefix

Carbon Flux analysis is used to produce statistics from sensitivity analysis runs of the forest carbon flux model. It uses the same approach as the Carbon Flux Full Standard Model, but with fewer model outputs analyzed. The sensitivity analysis argument needs to match the folder names in the input tile folders (e.g., maxgain, biomass_swap). To run this model with the standard flux model output for an analysis using fewer flux model outputs, use standard as the sensitivity_type argument. It currently only works with GADM features.

sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --analysis carbon_sensitivity --feature_type gadm --tcl --features s3://bucket/prefix/file.tsv --output s3://bucket/prefix --sensitivity_type sensitivity_analysis_type
sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --analysis carbon_sensitivity --feature_type gadm --tcl --features s3://bucket/prefix/file.tsv --output s3://bucket/prefix --sensitivity_type standard

Glad alert analysis computes whitelist, summary, daily and weekly change data for given input features and intersects areas with the same contextual layers as in annual update minimal. It is used to update the country and user dashboards for the GFW website.

Users can select, if they want to run the full analysis, or only look at change data. Computing only change data makes sense, if neither input features, nor contextual layers have changed, but only glad alerts. In that case, only the daily and weekly change tables will be updated.

Supported input features are

• GADM
• Geostore
• WDPA
• Simple Feature

sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --analysis gladalerts --feature_type gadm --glad --features s3://bucket/prefix/file.tsv --output s3://bucket/prefix [--change_only]
sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --analysis gladalerts --feature_type wdpa --glad --features s3://bucket/prefix/file.tsv --output s3://bucket/prefix [--change_only]
sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --analysis gladalerts --feature_type geostore --glad --features s3://bucket/prefix/file.tsv --output s3://bucket/prefix [--change_only]
sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --analysis gladalerts --feature_type feature --glad --features s3://bucket/prefix/file.tsv --output s3://bucket/prefix [--change_only]

Use Polygon Features encoded in TSV format. Geometries must be encoded in WKB. You can specify one or many input files using wildcards:

ex:

• s3://bucket/prefix/gadm36_1_1.tsv
• s3://bucket/prefix/geostore_*.tsv

Make sure features are sufficiently small to assure a well balanced partition size and workload. Larger features should be split into smaller features, prior to running the analysis. Also make sure, that features do not overlap with tile boundaries (we use 10x10 degree tiles). For best performance, intersect input features with a 1x1 degree grid. If you are not sure how to best approach this, simply use the ArcPY Client

The following options are supported:

• build.sbt: Scala Build Tool build configuration file
• .sbtopts: Command line options for SBT, including JVM parameters
• project: Additional configuration for SBT project, plugins, utility, versions
• src/main/scala: Application and utility code
• src/test/scala: Unit test files

For local testing input should be limited with --limit flag to minimize the time.

sbt:geotrellis-wri> test:runMain org.globalforestwatch.summarystats.SummaryMain --features file:/Users/input/ten-by-ten-gadm36/wdpa__10N_010E.tsv --output file:/User/out/summary --limit 10

Before running review sbtlighter configuration in build.sbt, reload SBT session if modified.

sbt:geotrellis-wri> sparkCreateCluster
sbt:treecoverloss> sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain --features s3://gfw-files/2018_update/tsv/gadm36_1_1.csv --output s3://gfw-files/2018_update/results/summary --feature_type gadm --analysis annualupdate_minimal --tcl
sbt:treecoverloss> sparkSubmitMain org.globalforestwatch.summarystats.SummaryMain  --features s3://gfw-files/2018_update/tsv/wdpa__*.tsv --output s3://gfw-files/2018_update/results/summary  --feature_type wdpa --analysis gladalerts --tcl --iso BRA