Skip to article frontmatterSkip to article content
Site not loading correctly?

This may be due to an incorrect BASE_URL configuration. See the MyST Documentation for reference.

Sentinel-2 power spectrum on equal-area HEALPix

Lifewatch ERIC

A small, reproducible GRID4EARTH example that exercises the analysis side of the stack: take a real Sentinel-2 L2A scene, put it on an equal-area HEALPix grid on the WGS84 ellipsoid with healpix-geo, and compute its isotropic power spectrum with healpix-analyse — a differentiable (PyTorch) toolkit for analysing signals on HEALPix grids.

Why it matters

The power spectrum measures at which spatial scales a field has structure. To be meaningful it must be computed on cells of equal surface area — a naïve lon/lat grid shrinks cells towards the poles and biases the spectrum. HEALPix cells all cover the same area at a given depth (nside = 2**depth), so healpix-geo gives an unbiased grid and healpix-analyse does the analysis on it. And because healpix-analyse is built on PyTorch, every operator is differentiable — the same pipeline drops straight into a learning loop.

What the notebook does

The angular power spectrum of a Sentinel-2 scene on equal-area HEALPix

  1. reads a window of Sentinel-2 L2A band B04 straight from the public AWS Cloud-Optimized GeoTIFF archive (no credentials);

  2. samples it onto a HEALPix rectangle at depth 18 (~25 m cells, WGS84) with healpix-geo;

  3. computes the radially-averaged power spectrum with healpix-analyse.

Self-contained and runs in seconds — the Jupyter Book builds it on every push.

Reproduce it

Sentinel-2 power spectrum on equal-area HEALPix
The angular power spectrum of a Sentinel-2 scene on equal-area HEALPix