This Jupyter Book presents the results of a replication study of the benchmarks from:
Law, R.M. & Ardo, J. (2024). Using a discrete global grid system for a scalable, interoperable, and reproducible system of land-use mapping. Big Earth Data, DOI: Law & Ardo (2024)
Original benchmark code: dggsBenchmarks v1.1.1.
The replication has three layers:
Reproduction — same methodology, same tools (H3 + Polars).
Replication — same methodology, alternative tools (
xdggs).Extension (v3.0.0) — HEALPix benchmarks on the sphere and on the WGS84 ellipsoid via the
healpix-geolibrary.
from _helpers import ROOT, load_json
print(f"Repository root: {ROOT}")Repository root: /home/runner/work/dggs_replication_2026/dggs_replication_2026
Run environment¶
The committed results record the exact environment they were produced in.
info = load_json("results_h3/system_info.json")
print(f"Code version : {info['code_version']}")
print(f"Timestamp : {info['timestamp']}")
print(f"Python : {info['python_version']}")
print(f"CPU count : {info['system']['cpu_count']}")
print(f"Memory (GB) : {info['system']['memory_gb']}")
print()
print("Pinned dependencies:")
for pkg, ver in info["dependencies"].items():
print(f" {pkg:<12} {ver}")Code version : 2026-01-20-unified-v3
Timestamp : 2026-03-07T19:17:55.589337
Python : 3.12.12
CPU count : 10
Memory (GB) : 32.0
Pinned dependencies:
numpy 2.4.1
pandas 2.3.3
geopandas 1.1.2
h3 4.4.1
xdggs available
polars 1.37.1
Benchmark configuration¶
cfg = info["configuration"]
for key, val in cfg.items():
print(f"{key:<16}: {val}")vector_layers : [5, 10, 20, 50]
raster_layers : [10, 50, 100, 500]
h3_resolution : 9
random_seed : 42
Paper claims under test¶
The two central performance claims of Law & Ardo (2024):
summary = load_json("results_h3/summary.json")
for name, claim in summary["paper"]["claims"].items():
print(f"- {name.capitalize()}: {claim}")- Vector: DGGS provides orders of magnitude performance improvement
- Raster: DGGS and raster methods show roughly equivalent performance
- Law, R. M., & Ardo, J. (2024). Using a discrete global grid system for a scalable, interoperable, and reproducible system of land-use mapping. Big Earth Data, 9(1), 29–46. 10.1080/20964471.2024.2429847