dggs-biodiversity-bias — why equal-area cells matter for biodiversity (and why DGGS adds shape and hierarchy)

Supporting evidence for the EGU 2026 talk EGU26-11348 (Fouilloux et al., LifeWatch ERIC, ESSI2.6, 3–8 May 2026).

This release expands the v0.1.0 evidence base from a 3-grid comparison (lat-lon vs Behrmann vs HEALPix) to a comprehensive 7-grid evaluation across every grid family the biodiversity community uses in practice, and isolates HEALPix-specific advantages for the integration of biodiversity science with Copernicus Earth-observation data and Destination Earth climate models.

The eight-notebook argument

1. Equal-area is necessary. Lat-lon over-counts biodiversity by up to 23× at 5° resolution, purely from cell geometry (notebooks 01–02).
2. Six equal-area choices pass the count test. HEALPix, H3, rHEALPix, ISEA3H, Mollweide, and the EEA reference grid (LAEA Europe / EPSG:3035, the INSPIRE / Habitats Directive standard) all agree on biodiversity density patterns over Quercus suber's Mediterranean range. For biodiversity counts in isolation, the choice between them is not a count-correctness question (notebook 07).
3. DGGS family preserves cell shape across latitudes; projection family does not. Behrmann, Mollweide, and the EEA reference grid distort poleward at high latitudes; HEALPix, H3, rHEALPix, and ISEA3H all preserve compact cells everywhere — the property that makes ML kernels mean the same geographic operator on every part of Earth (notebooks 03–04).
4. HEALPix is the right common DGGS for the integrated future of biodiversity science with high-resolution Copernicus EO and Destination Earth climate models, by virtue of: native geometric deep learning on the sphere (DeepSphere, spherical CNNs); scattering networks via foscat (the FIESTA stack); native sphere-harmonic transforms (healpy.map2alm / alm2map); NESTED bit-shift hierarchical refinement (zoom is O(1) per cell); iso-latitude pixelization for zonal climate-zone analyses; and a credible ellipsoidally-correct path via rHEALPix or "Ellipsoidal HEALPix" (the ESA GRID4EARTH approach). For pure biodiversity-only work at coarse resolution, sphere HEALPix is fine; for the integration regime, ellipsoidal correctness becomes material (notebooks 06, 08).

Connection to ESA GRID4EARTH

This Jupyter Book is the biodiversity-side version of the case ESA GRID4EARTH makes for Ellipsoidal HEALPix as a Common DGGS for Copernicus EO and Destination Earth — bridging spherical climate models and ellipsoidal Earth-observation data on a single ellipsoidally-correct, hierarchical, scalable DGGS.

Cite

If you use this work please cite this release (DOI minted by Zenodo upon publication) and the foundational references in `CITATION.cff`: Górski et al. 2005 (HEALPix), Sahr et al. 2003 (DGGS), Hauffe et al. 2023 (DGGS for biodiversity), Kmoch et al. 2022 (DGGS area distortions).