Skip to content

Jupyter Notebook Example

This page shows a complete Jupyter-style workflow using cross-sensor-cal:

  1. Run the pipeline from Python
  2. Preview the merged Parquet table
  3. Compute a simple NDVI diagnostic

You can copy these cells directly into a notebook.

1. Imports and configuration

```python import os import duckdb import pandas as pd

from cross_sensor_cal import go_forth_and_multiply

base_folder = "output_notebook_demo" site_code = "NIWO" year_month = "2023-08" product_code = "DP1.30006.001" flight_lines = ["NEON_D13_NIWO_DP1_L020-1_20230815_directional_reflectance"] 2. Run the pipeline from Python go_forth_and_multiply( base_folder=base_folder, site_code=site_code, year_month=year_month, product_code=product_code, flight_lines=flight_lines, max_workers=2, engine="thread", # start with thread; switch to "ray" if configured ) This cell will: download NEON HDF5 tiles if needed export ENVI cubes apply topographic + BRDF correction convolve to Landsat (if configured) write Parquet tables generate QA PNG, PDF, and JSON 3. Locate the merged Parquet file fl_prefix = flight_lines[0] fl_dir = os.path.join(base_folder, fl_prefix)

merged = os.path.join( fl_dir, f"{fl_prefix}_merged_pixel_extraction.parquet", ) merged 4. Preview the merged table with DuckDB con = duckdb.connect()

head_df = con.execute( f"SELECT * FROM read_parquet('{merged}') LIMIT 5" ).df() head_df 5. Compute a quick NDVI diagnostic This example assumes Landsat-style band names Red and NIR are present in the merged table. ndvi_df = con.execute( """ SELECT Red, NIR, (NIR - Red) / NULLIF(NIR + Red, 0) AS ndvi FROM read_parquet($merged) LIMIT 5000 """, {"merged": merged}, ).df()

ndvi_df["ndvi"].describe() You can extend this pattern to: filter by masks aggregate by plot or polygon export subsets for modeling 6. Visualize NDVI distribution (optional) If you have matplotlib installed: import matplotlib.pyplot as plt

ndvi_df["ndvi"].plot.hist(bins=50) plt.xlabel("NDVI") plt.ylabel("Count") plt.title("NDVI distribution (sample)") plt.show() Next steps Quickstart Working with Parquet outputs Pipeline overview & stages