Sentinel-2 Analysis Notebook

Interactive Jupyter notebook for comprehensive Sentinel-2 data analysis.

Notebook Overview

This notebook demonstrates a complete workflow for Sentinel-2 satellite imagery analysis, including:

  • Data discovery and filtering

  • Multi-temporal analysis

  • NDVI calculation and visualization

  • Change detection

  • Export and reporting

Prerequisites

pip install open-geodata-api[complete] jupyter matplotlib seaborn plotly folium

Notebook Sections

1. Setup and Configuration

import open_geodata_api as ogapi
import rioxarray as rxr
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np
from open_geodata_api.utils import filter_by_cloud_cover

# Setup
pc = ogapi.planetary_computer(auto_sign=True)
study_area = [-122.5, 47.5, -122.0, 48.0]  # Seattle area

2. Data Discovery

# Search for Sentinel-2 data
results = pc.search(
    collections=['sentinel-2-l2a'],
    bbox=study_area,
    datetime='2024-01-01/2024-12-31',
    query={'eo:cloud_cover': {'lt': 50}},
    limit=50
)

items = results.get_all_items()
print(f"Found {len(items)} Sentinel-2 scenes")

# Convert to DataFrame for analysis
df = items.to_dataframe()
df.head()

3. Quality Assessment and Filtering

# Cloud cover analysis
plt.figure(figsize=(12, 4))

plt.subplot(1, 2, 1)
plt.hist(df['eo:cloud_cover'], bins=20, alpha=0.7)
plt.xlabel('Cloud Cover (%)')
plt.ylabel('Number of Scenes')
plt.title('Cloud Cover Distribution')

plt.subplot(1, 2, 2)
df['date'] = pd.to_datetime(df['datetime'])
plt.scatter(df['date'], df['eo:cloud_cover'], alpha=0.6)
plt.xlabel('Date')
plt.ylabel('Cloud Cover (%)')
plt.title('Cloud Cover Over Time')
plt.xticks(rotation=45)

plt.tight_layout()
plt.show()

4. Multi-temporal NDVI Analysis

# Filter for clear images
clear_items = filter_by_cloud_cover(items, max_cloud_cover=20)

# Calculate NDVI time series
ndvi_time_series = []

for item in clear_items[:12]:  # Process first 12 clear scenes
    try:
        # Get NIR and Red bands
        urls = item.get_band_urls(['B08', 'B04'])

        nir = rxr.open_rasterio(urls['B08'])
        red = rxr.open_rasterio(urls['B04'])

        # Calculate NDVI
        ndvi = (nir - red) / (nir + red)

        # Calculate statistics
        ndvi_time_series.append({
            'date': item.properties['datetime'][:10],
            'mean_ndvi': float(ndvi.mean()),
            'std_ndvi': float(ndvi.std()),
            'cloud_cover': item.properties.get('eo:cloud_cover', 0)
        })

    except Exception as e:
        print(f"Error processing {item.id}: {e}")

5. Visualization and Analysis

# Create comprehensive visualization
ndvi_df = pd.DataFrame(ndvi_time_series)
ndvi_df['date'] = pd.to_datetime(ndvi_df['date'])

fig, ((ax1, ax2), (ax3, ax4)) = plt.subplots(2, 2, figsize=(15, 10))

# NDVI time series
ax1.plot(ndvi_df['date'], ndvi_df['mean_ndvi'], 'o-', color='green')
ax1.set_title('NDVI Time Series')
ax1.set_ylabel('Mean NDVI')
ax1.tick_params(axis='x', rotation=45)

# NDVI distribution
ax2.hist(ndvi_df['mean_ndvi'], bins=15, alpha=0.7, color='green')
ax2.set_title('NDVI Distribution')
ax2.set_xlabel('NDVI')
ax2.set_ylabel('Frequency')

# NDVI vs Cloud Cover
ax3.scatter(ndvi_df['cloud_cover'], ndvi_df['mean_ndvi'], alpha=0.7)
ax3.set_title('NDVI vs Cloud Cover')
ax3.set_xlabel('Cloud Cover (%)')
ax3.set_ylabel('Mean NDVI')

# Seasonal patterns
ndvi_df['month'] = ndvi_df['date'].dt.month
monthly_ndvi = ndvi_df.groupby('month')['mean_ndvi'].mean()
ax4.plot(monthly_ndvi.index, monthly_ndvi.values, 'o-', color='darkgreen')
ax4.set_title('Seasonal NDVI Pattern')
ax4.set_xlabel('Month')
ax4.set_ylabel('Mean NDVI')
ax4.set_xticks(range(1, 13))

plt.tight_layout()
plt.show()

6. Export and Reporting

# Create summary report
summary = {
    'analysis_period': f"{ndvi_df['date'].min()} to {ndvi_df['date'].max()}",
    'total_scenes': len(ndvi_df),
    'mean_ndvi': ndvi_df['mean_ndvi'].mean(),
    'ndvi_range': [ndvi_df['mean_ndvi'].min(), ndvi_df['mean_ndvi'].max()],
    'seasonal_variation': monthly_ndvi.max() - monthly_ndvi.min()
}

print("=== Sentinel-2 NDVI Analysis Summary ===")
for key, value in summary.items():
    print(f"{key}: {value}")

# Export data
ndvi_df.to_csv('sentinel2_ndvi_analysis.csv', index=False)
print("\nData exported to: sentinel2_ndvi_analysis.csv")

Download and Run

  1. Download the complete notebook:

    Visit the examples repository and download sentinel2_analysis.ipynb

  2. Start Jupyter:

jupyter notebook sentinel2_analysis.ipynb

  1. Run cells sequentially and modify parameters as needed

Key Learning Outcomes

After completing this notebook, you will understand:

βœ… Data Discovery - How to find and filter Sentinel-2 imagery βœ… Quality Assessment - Evaluating data quality using cloud cover βœ… Multi-temporal Analysis - Working with time series satellite data βœ… NDVI Calculation - Computing vegetation indices from spectral bands βœ… Visualization - Creating informative plots and charts βœ… Export Workflows - Saving results for further analysis

Extensions

Try these extensions to deepen your understanding:

  • Change Detection: Compare NDVI between two time periods

  • Spatial Analysis: Focus on specific land cover types

  • Cross-sensor Comparison: Compare with Landsat data

  • Advanced Indices: Calculate EVI, SAVI, or other vegetation indices

  • Interactive Maps: Use Folium for interactive visualizations