NetCDF Territorial Classification and Python Colab Scripts Based on Welch Power Spectrum SbAI Signals at 1 Cycle per Year in Southern South America (Argentina, Chile, and Uruguay). (FS-WEL-1y)
Otros responsables
Fecha
Publicador:
Consejo Nacional de Investigaciones Científicas y Técnicas
Fecha de depósito:
27/05/2026
Fecha de creación:
14/5/2026
Clasificación temática
Palabras clave
Resumen
This notebook implements a full pipeline for band-pass spectral analysis and clustering of FIR-filtered SbAI data:
- Applies FIR band-pass filter (0.9–1.1 cycles/year, ~1 year).
- Computes Welch spectra per pixel in the 1-year band.
- Calculates preserved energy maps with provincial boundaries.
- Applies RobustScaler preprocessing to mitigate outliers.
- Performs KMeans++ clustering (n_clusters=5, n_init=10).
- Generates classification maps with geographic boundaries from a geopackage.
- Produces cluster centroids across the 1-year band and representative pixels closest to centroids.
- Provides autonomous visualization scripts to replicate maps and spectral plots without recalculating clustering.
Información Técnica
The workflow consists of:
1. Mounting Google Drive and setting the working directory (`base_dir`).
2. Loading raw dataset (`netcdf_sbai_025_raw.nc`).
3. Applying FIR band-pass filter (0.9–1.1 cpa, ~1 year).
4. Saving filtered dataset (`netcdf_0.25_bandpass_1y.nc`).
5. Computing Welch spectra per pixel:
- Frequency resolution increased with `nfft`.
- Interpolating spectra to include exact band edges (0.9 and 1.1 cpa).
- Saving results as NetCDF (`welch_fft_1y.nc`).
6. Running KMeans++ clustering (5 clusters, RobustScaler, n_init=10):
- Computing metrics: explained variance (R²), silhouette index, relative reconstruction error.
- Filtering clusters below minimum threshold (1% of pixels).
- Saving classification results (`clasificacion_kmeans_1y.nc`).
7. Plotting:
- Geographic cluster map with `canvas.gpkg`.
- Spectral plots of cluster centroids.
- Spectral plots of representative pixels.
8. Autonomous visualization:
- Scripts to reproduce maps and spectral plots directly from saved NetCDFs.
- No need to rerun clustering.
Colecciones
Condiciones de uso
Las buenas prácticas científicas esperan que se otorgue el crédito adecuado mediante una citación. Utilice un formato de citación y aplique estas normas de reutilización:
Licencia Creative Commons
Excepto donde se indique lo contrario, la licencia de este ítem se describe como by-nc-sa
Archivos
ARCHIVO
NOTAS DE USO
TAMAÑO