sm-global-v2 (NDVI AOI timeseries integrated from your notebook)

This repo wraps your NDVI AOI timeseries notebook code into a reusable module and CLI, adds DPS YAML for scale-out, and includes stubs for your 3-stage SM pipeline.

Install (in MAAP ADE)

mamba env update -f env.yml
conda activate sm-global-env

Run (point-based SMAP 9km cell AOI)

python algos/ndvi_aoi_timeseries.py \  --lon -109.35661 --lat 32.03021 \  --start 2025-02-01 --end 2025-09-01 \  --pix-m 1000 \  --limit 100 \  --out-dir ./outputs

Outputs:

• outputs/ndvi_ts_<start>_<end>_<pix>m.csv and .parquet with columns: date, sensor, tile, url_s3, ndvi_mean, coverage_frac, n_valid, n_total

DPS registration

• Use dps/ndvi-aoi-timeseries.yml to register the NDVI CLI in MAAP.
• The three SM pipeline stubs and their DPS YAMLs are also included under algos/ and dps/.

Notes

• Uses ease-lonlat for EASE2 indexing & corners.
• Uses earthaccess for LP DAAC search/credentials and robust S3/HTTPS opening.
• Resampling uses a grid derived from the first scene's CRS to ensure array compatibility across scenes.
• To switch to a 200 m EASE2 target instead of UTM, swap target_grid_from_aoi with an EASE2-based window function.