Filtering an ImageCollection

  • Earth Engine provides multiple methods for filtering image collections, including convenience functions like filterDate() and filterBounds() as well as the more general filter() method for custom filtering needs.

  • This example demonstrates how to filter a Landsat 8 image collection by date, month, geographic bounds, and cloud cover using these methods.

  • Filtering by cloud cover significantly improves the quality of composites derived from image collections, as shown by comparing a composite generated from unfiltered data with one generated from data filtered for zero cloud cover.

  • The code example is provided in both JavaScript and Python, enabling users to apply these filtering techniques in their preferred programming environment within the Earth Engine platform.

As illustrated in the Get Started section and the ImageCollection Information section, Earth Engine provides a variety of convenience methods for filtering image collections. Specifically, many common use cases are handled by imageCollection.filterDate(), and imageCollection.filterBounds(). For general purpose filtering, use imageCollection.filter() with an ee.Filter as an argument. The following example demonstrates both convenience methods and filter() to identify and remove images with high cloud cover from an ImageCollection.

Code Editor (JavaScript)

// Load Landsat 8 data, filter by date, month, and bounds.
var collection = ee.ImageCollection('LANDSAT/LC08/C02/T1_TOA')
  .filterDate('2015-01-01', '2018-01-01')  // Three years of data
  .filter(ee.Filter.calendarRange(11, 2, 'month'))  // Only Nov-Feb observations
  .filterBounds(ee.Geometry.Point(25.8544, -18.08874));  // Intersecting ROI

// Also filter the collection by the CLOUD_COVER property.
var filtered = collection.filter(ee.Filter.eq('CLOUD_COVER', 0));

// Create two composites to check the effect of filtering by CLOUD_COVER.
var badComposite = collection.mean();
var goodComposite = filtered.mean();

// Display the composites.
Map.setCenter(25.8544, -18.08874, 13);
Map.addLayer(badComposite,
             {bands: ['B3', 'B2', 'B1'], min: 0.05, max: 0.35, gamma: 1.1},
             'Bad composite');
Map.addLayer(goodComposite,
             {bands: ['B3', 'B2', 'B1'], min: 0.05, max: 0.35, gamma: 1.1},
             'Good composite');

Python setup

See the Python Environment page for information on the Python API and using geemap for interactive development.

import ee
import geemap.core as geemap

Colab (Python)

# Load Landsat 8 data, filter by date, month, and bounds.
collection = (
    ee.ImageCollection('LANDSAT/LC08/C02/T1_TOA')
    # Three years of data
    .filterDate('2015-01-01', '2018-01-01')
    # Only Nov-Feb observations
    .filter(ee.Filter.calendarRange(11, 2, 'month'))
    # Intersecting ROI
    .filterBounds(ee.Geometry.Point(25.8544, -18.08874))
)

# Also filter the collection by the CLOUD_COVER property.
filtered = collection.filter(ee.Filter.eq('CLOUD_COVER', 0))

# Create two composites to check the effect of filtering by CLOUD_COVER.
bad_composite = collection.mean()
good_composite = filtered.mean()

# Display the composites.
m = geemap.Map()
m.set_center(25.8544, -18.08874, 13)
m.add_layer(
    bad_composite,
    {'bands': ['B3', 'B2', 'B1'], 'min': 0.05, 'max': 0.35, 'gamma': 1.1},
    'Bad composite',
)
m.add_layer(
    good_composite,
    {'bands': ['B3', 'B2', 'B1'], 'min': 0.05, 'max': 0.35, 'gamma': 1.1},
    'Good composite',
)
m