POD fitting on real satellite data — placeholder

POD fitting on real satellite data¶

This slot is reserved for the real-data companion to 06_stationary_numpyro_mcmc. The original james_jax.py script reads the IMEO validated-plumes CSV + the Tanager plumes CSV, computes the $Q/U$ proxy per satellite, and runs a per-satellite NUTS fit using the same pod_powerlaw_model that we fit on synthetic data in notebook 06.

The library code for this is already in place: see methane_pod.fitting.run_mcmc, pod_powerlaw_model, lognorm_cdf, power_law. What is missing is the CSV ingestion layer (satellite-specific filters, Q/U computation, outlier handling) and the source data itself.

When the datasets land, the notebook will:

1. Load IMEO + Tanager plume tables.
2. Filter to Oil & Gas, apply per-satellite outlier thresholds.
3. Fit the POD-modified power law independently per satellite.
4. Overlay recovered POD curves with ±1σ bands on a single axis for a head-to-head platform comparison.
5. Compare the fitted $x_{50}$ and σ to published values (Varon 2018, Cusworth 2021, Kamdar IMEO).