Satellite Imagery Analysis - Spectral Indices Explained

Understanding Spectral Indices

Spectral indices are mathematical combinations of spectral bands that enhance specific features in satellite imagery. By leveraging the unique reflectance properties of different materials, these indices help us quantify vegetation health, water content, urban density, and more.

The power of spectral indices lies in their simplicity and effectiveness. A single index value can tell us more about a pixel than examining individual bands separately.

Vegetation Indices

NDVI - Normalized Difference Vegetation Index

The Formula:

NDVI is the most widely used vegetation index, exploiting the fact that healthy vegetation:

• Absorbs red light (for photosynthesis)
• Reflects near-infrared light (due to leaf cell structure)

Value Interpretation:

• Agriculture
• Forestry
• Environmental

• Crop Health Monitoring

  # Calculate NDVI using Python
  import rasterio
  import numpy as np
  
  # Read NIR (Band 8) and Red (Band 4) from Sentinel-2
  with rasterio.open('B08.tif') as nir:
      nir_band = nir.read(1).astype(float)
      
  with rasterio.open('B04.tif') as red:
      red_band = red.read(1).astype(float)
  
  # Calculate NDVI
  ndvi = (nir_band - red_band) / (nir_band + red_band)
  
  # Mask out non-vegetation areas
  ndvi_vegetation = np.where(ndvi > 0.2, ndvi, np.nan)
  

  Use Cases:

  • Identify stressed crops before visible symptoms
  • Optimize variable-rate fertilizer application
  • Estimate biomass and yield potential
  • Monitor crop phenology throughout growing season

• Forest Health Assessment

  NDVI time-series analysis reveals:

  • Deforestation events: Sharp drops in NDVI values
  • Seasonal patterns: Deciduous vs. evergreen forests
  • Pest infestations: Gradual NDVI decline over weeks/months
  • Fire damage: Immediate NDVI reduction in burned areas

  Example Workflow:

  1. Download 5 years of Sentinel-2 imagery
  2. Calculate NDVI for each image
  3. Create median composite for each season
  4. Detect anomalies using statistical thresholds
  5. Generate alerts for areas with >20% NDVI decrease

• Ecosystem Monitoring

  • Wetland mapping: NDVI helps distinguish vegetated wetlands from open water
  • Desertification tracking: Monitor vegetation loss in arid regions
  • Carbon sequestration: NDVI correlates with biomass and carbon storage
  • Biodiversity assessment: Vegetation structure indicates habitat quality

Seasonal NDVI Trends

This chart shows typical NDVI patterns for different vegetation types throughout the year:

{
  "type": "line",
  "data": {
    "labels": ["Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"],
    "datasets": [
      {
        "label": "Deciduous Forest (Temperate)",
        "data": [0.25, 0.28, 0.45, 0.68, 0.78, 0.82, 0.80, 0.75, 0.65, 0.48, 0.32, 0.26],
        "borderColor": "#27ae60",
        "backgroundColor": "rgba(39, 174, 96, 0.1)",
        "fill": true,
        "tension": 0.4
      },
      {
        "label": "Evergreen Forest (Tropical)",
        "data": [0.75, 0.76, 0.77, 0.78, 0.79, 0.80, 0.81, 0.80, 0.79, 0.78, 0.77, 0.76],
        "borderColor": "#229954",
        "backgroundColor": "rgba(34, 153, 84, 0.1)",
        "fill": true,
        "tension": 0.4
      },
      {
        "label": "Wheat (Northern Hemisphere)",
        "data": [0.15, 0.18, 0.35, 0.55, 0.72, 0.78, 0.45, 0.22, 0.15, 0.15, 0.15, 0.15],
        "borderColor": "#f39c12",
        "backgroundColor": "rgba(243, 156, 18, 0.1)",
        "fill": true,
        "tension": 0.4
      },
      {
        "label": "Grassland (Semi-Arid)",
        "data": [0.20, 0.22, 0.28, 0.42, 0.55, 0.48, 0.38, 0.32, 0.28, 0.25, 0.22, 0.20],
        "borderColor": "#d4ac0d",
        "backgroundColor": "rgba(212, 172, 13, 0.1)",
        "fill": true,
        "tension": 0.4
      }
    ]
  },
  "options": {
    "responsive": true,
    "plugins": {
      "title": {
        "display": true,
        "text": "Seasonal NDVI Patterns by Vegetation Type",
        "font": {
          "size": 16
        }
      },
      "legend": {
        "position": "bottom"
      }
    },
    "scales": {
      "y": {
        "beginAtZero": true,
        "max": 1.0,
        "title": {
          "display": true,
          "text": "NDVI Value"
        }
      },
      "x": {
        "title": {
          "display": true,
          "text": "Month"
        }
      }
    }
  }
}

Water Indices

NDWI - Normalized Difference Water Index

The Formula:

NDWI enhances water features by exploiting water’s high absorption in NIR wavelengths.

Alternative SWIR-based NDWI:

This version is better for vegetation water content assessment.

• Flood Mapping
• Water Quality
• Irrigation

• Rapid Flood Assessment

  # Calculate NDWI for flood detection
  import rasterio
  import numpy as np
  
  # Read Green (Band 3) and NIR (Band 8)
  with rasterio.open('B03.tif') as green:
      green_band = green.read(1).astype(float)
      
  with rasterio.open('B08.tif') as nir:
      nir_band = nir.read(1).astype(float)
  
  # Calculate NDWI
  ndwi = (green_band - nir_band) / (green_band + nir_band)
  
  # Threshold for water detection
  water_mask = ndwi > 0.3
  

  Applications:

  • Emergency response during flood events
  • Compare pre-flood and post-flood imagery
  • Estimate affected area and population
  • Monitor water recession over time

• Lake and Reservoir Monitoring

  NDWI combined with other indices reveals:

  • Turbidity levels: Sediment-laden water has different NDWI
  • Algal blooms: Chlorophyll affects green band reflectance
  • Water extent changes: Drought monitoring in reservoirs
  • Wetland dynamics: Seasonal water level fluctuations

  Monitoring Workflow:

  1. Monthly NDWI calculation for water body
  2. Extract water extent polygon
  3. Calculate area and compare to historical baseline
  4. Generate alerts if area drops below threshold

• Agricultural Water Management

  SWIR-based NDWI for crop water stress:

  • Values > 0.5: Well-watered vegetation
  • Values 0.2-0.5: Moderate water stress
  • Values < 0.2: Severe water stress

  Helps optimize irrigation scheduling and detect leaks in irrigation systems.

Urban and Built-up Indices

NDBI - Normalized Difference Built-up Index

The Formula:

NDBI highlights built-up areas, as urban surfaces reflect more SWIR than NIR.

UI - Urban Index

The Formula:

Similar to NDBI but uses SWIR2 band for better discrimination.

Comprehensive Index Comparison

Multi-Year NDVI Analysis

This interactive visualization shows NDVI trends for a hypothetical agricultural region over 5 years:

{
  "$schema": "https://vega.github.io/schema/vega-lite/v5.json",
  "description": "Multi-year NDVI trends showing seasonal patterns and long-term changes",
  "width": 700,
  "height": 400,
  "data": {
    "values": [
      {"date": "2019-01", "ndvi": 0.35, "year": "2019"},
      {"date": "2019-04", "ndvi": 0.62, "year": "2019"},
      {"date": "2019-07", "ndvi": 0.78, "year": "2019"},
      {"date": "2019-10", "ndvi": 0.48, "year": "2019"},
      {"date": "2020-01", "ndvi": 0.32, "year": "2020"},
      {"date": "2020-04", "ndvi": 0.58, "year": "2020"},
      {"date": "2020-07", "ndvi": 0.72, "year": "2020"},
      {"date": "2020-10", "ndvi": 0.45, "year": "2020"},
      {"date": "2021-01", "ndvi": 0.28, "year": "2021"},
      {"date": "2021-04", "ndvi": 0.55, "year": "2021"},
      {"date": "2021-07", "ndvi": 0.68, "year": "2021"},
      {"date": "2021-10", "ndvi": 0.42, "year": "2021"},
      {"date": "2022-01", "ndvi": 0.38, "year": "2022"},
      {"date": "2022-04", "ndvi": 0.65, "year": "2022"},
      {"date": "2022-07", "ndvi": 0.82, "year": "2022"},
      {"date": "2022-10", "ndvi": 0.52, "year": "2022"},
      {"date": "2023-01", "ndvi": 0.40, "year": "2023"},
      {"date": "2023-04", "ndvi": 0.68, "year": "2023"},
      {"date": "2023-07", "ndvi": 0.85, "year": "2023"},
      {"date": "2023-10", "ndvi": 0.55, "year": "2023"}
    ]
  },
  "mark": {
    "type": "line",
    "point": true,
    "tooltip": true
  },
  "encoding": {
    "x": {
      "field": "date",
      "type": "temporal",
      "title": "Date",
      "axis": {
        "format": "%Y-%m"
      }
    },
    "y": {
      "field": "ndvi",
      "type": "quantitative",
      "title": "NDVI Value",
      "scale": {
        "domain": [0, 1]
      }
    },
    "color": {
      "field": "year",
      "type": "nominal",
      "title": "Year",
      "scale": {
        "scheme": "category10"
      }
    }
  }
}

Interpretation: This data shows improving vegetation health over the 5-year period, with peak NDVI values increasing from 0.78 (2019) to 0.85 (2023), suggesting successful agricultural practices or favorable climate conditions.

Best Practices

1. Choose the Right Index

• Dense vegetation: Use EVI instead of NDVI to avoid saturation
• Sparse vegetation: Use SAVI with appropriate soil adjustment factor
• Water mapping: MNDWI performs better than NDWI in most cases
• Urban areas: Combine NDBI with NDVI for better discrimination

2. Account for Atmospheric Effects

Always apply atmospheric correction before calculating indices. Uncorrected imagery can lead to:

• Underestimated NDVI in hazy conditions
• False water detection due to cloud shadows
• Inconsistent time-series analysis

3. Validate with Ground Truth

• Collect field measurements when possible
• Compare index values with known land cover types
• Use high-resolution imagery for validation
• Consider seasonal variations in interpretation

4. Time-Series Analysis

Single-date indices have limitations. Multi-temporal analysis reveals:

• Phenological patterns
• Long-term trends
• Anomaly detection
• Change events

Practical Example: Drought Monitoring

Combining multiple indices provides comprehensive drought assessment:

1. NDVI: Detect vegetation stress
2. NDWI (SWIR): Measure vegetation water content
3. LST (Land Surface Temperature): Identify heat stress
4. Precipitation data: Correlate with index changes

Workflow:

# Pseudo-code for drought monitoring
for each_month in time_series:
    ndvi = calculate_ndvi(month)
    ndwi = calculate_ndwi_swir(month)
    
    # Drought indicator
    drought_index = (1 - ndvi) + (1 - ndwi)
    
    if drought_index > threshold:
        flag_as_drought_affected()

Conclusion

Spectral indices transform raw satellite imagery into actionable information. By understanding the principles behind each index and applying them appropriately, you can:

• Monitor vegetation health at field to continental scales
• Map water resources and track changes over time
• Identify urban expansion and infrastructure development
• Detect environmental changes and anomalies

In the next article, we’ll explore land cover classification using machine learning to automatically categorize every pixel in satellite imagery.

Questions about spectral indices? Want to see specific applications? Let me know!