SECTION 2 · FREQUENTLY ASKED

Common questions about the measurement.

Why do canopy figures from different sources differ?

Different definitions, different sensors, different vintages. A "canopy fraction" computed from NDVI at one resolution is not the same quantity as a "tree-cover" pixel class from Hansen GFC at a 30 percent threshold, or an "open-forest area" expressed in hectares. Leaves.PH publishes one specific quantity defined on the methodology page; adjacent published figures are listed as references with their respective definitions.

How does this measurement relate to DENR or GFW figures?

It is a separate measurement using its own definition (a human-calibrated canopy model over Sentinel-2 NDVI and spectral bands, Dynamic World tree probability, Meta v2 1 m canopy height, and ESA land cover, validated against manual high-resolution labels at F1 0.78). It is not a replacement for, or a verdict on, any other source. The methodology cross-reference on the home page lists adjacent published estimates side by side so the definitional differences are visible.

Does this prove a specific expressway or development clearing was illegal?

No. Leaves.PH publishes per-LGU canopy fractions from satellite imagery; it does not adjudicate Environmental Compliance Certificates, permits, or whether ECC conditions were met. Before-and-after canopy strips of specific corridors are descriptive evidence of vegetation change, not legal evidence of permit compliance.

Is there per-barangay data?

Yes. data/per_barangay/per_barangay_canopy_2019_2026.csv ships 892 OSM admin-level=10 polygons inside NCR, with canopy fractions per year per polygon. These are OSM subdivisions; mapping them onto the PSA barangay roster is the next data-engineering refinement.

Why 30 m resolution and not 10 m?

GEE's synchronous URL export limit is 50 MB. At 10 m native, a 4-band NCR Sentinel-2 composite is ~190 MB. At 30 m it fits inside the limit. The choice trades off some urban-tree detail for build simplicity (no chunked exports + mosaic step). The next granularity push may move to 10 m chunked.

What is the deal with the La Mesa watershed?

It is the NCR canopy story most people miss. Estoque et al. 2018 (Forest Ecology and Management vol. 430) documents a net forest gain of 557 hectares in La Mesa from 2002 to 2016 thanks to MWSI / Bantay Kalikasan reforestation. The decline narrative for Metro Manila is real but not monotone: Quezon City and Marikina hold the NE green zone and have been climbing while Manila and Caloocan continue to lose canopy.

Can I cite Leaves.PH in a paper?

Yes. See CITATION.cff. A versioned Zenodo DOI is minted at each tagged release.

How do I submit a correction?

Open an issue at github.com/xmpuspus/leaves-ph using the LGU correction template. Include the LGU name, the year, the alternative value you propose, and a citable source. Per-barangay corrections are welcome.

What if I find what looks like a private residence in the data?

Leaves.PH publishes canopy polygons (per-LGU, per-barangay, and per-crown) derived from satellite imagery and a 1 m canopy-height mask. Crown polygons outline vegetation, not buildings, and carry no household-level inference or individual identification. Vegetation is not personal data under RA 10173. If you still spot something concerning, file a private security advisory: github.com/xmpuspus/leaves-ph/security. Acknowledged within 5 working days.