Advanced Google Earth Engine(GEE) For Spatial Data Analysis

Explore the Google Earth Engine interface to access cloud-based, planetary-scale satellite imagery and geospatial data, run analyses with inbuilt algorithms, and review case studies and datasets.

Sign up for Google Earth Engine with a Gmail account to access the Earth Engine Code Editor and API, then learn to write code to access and visualize geospatial data.

Explore Google Earth Engine datasets, such as Landsat 5/7/8 composites and enhanced vegetation index, in the data catalog and workspace to analyze greenery and forest changes.

Explore the vast catalog of more than 700 Google Earth Engine datasets, from Sentinel and climate data to high‑resolution imagery, terrain, nightlights, and human modification layers.

Discover how to use the in-browser code editor for Google Earth Engine, create repositories and folders to organize scripts, save files, and begin with basic JavaScript concepts.

This whistle-stop tour introduces basic JavaScript concepts essential for using Google Earth Engine’s code editor, covering variables, strings, numbers, lists, dictionaries, comments, and inbuilt functions.

Adapt JavaScript for Google Earth Engine by declaring variables and data types such as feature, image, and image collection; import assets, print contents, and visualize with map.

learn how to upload your own data to Google Earth Engine by ingesting shapefiles or individual files as assets, monitor ingest progress, handle errors, and understand projection caveats.

Learn how shapefiles and feature collections organize points, polygons, and lines in Google Earth Engine, including importing external data and manually defining custom geometries.

Filter shapefiles in Google Earth Engine by country name or code to zoom into a specific country, using a world feature collection and selecting Thailand and Cambodia.

Filter multiple countries in Google Earth Engine by creating a list of two-letter country codes and applying in list and dot filter to a feature collection.

Learn to create a buffer around a feature collection by mapping a buffer function across points, lines, or polygons with an adjustable distance. See how to style and visualize the result.

Compute zonal statistics on feature data to summarize protected areas in Malaysia by IUCN designation, yielding sums and averages for representative and total areas.

Explore intersecting overlapping feature collections in Google Earth Engine by filtering the ecoregions global dataset to Thailand, defining a custom intersect function, and applying symbology.

Learn to read and visualize a single-band digital elevation model in Google Earth Engine, using one-band data, map centering, zoom, and custom visualization with color palettes.

Explore visualizing imagery data and image collections in Google Earth Engine by selecting the land cover band and mapping six classes (0–5) with a six color palette and visualization parameters.

Explore a global digital elevation model and apply conditions to classify terrain into three categories—less than 500 m, 500–1000 m, and above 1000 m—and visualize as a three-color map.

Derive slope and aspect from a digital elevation model in Google Earth Engine, computing slope as the maximum rate of change and aspect as slope direction.

Explore resampling in Google Earth Engine, where nearest-neighbor resampling is the default for raster data, and you can switch to cubic resampling for crisper grayscale digital elevation models.

Explore image collections in Google Earth Engine, including Landsat 8 and Sentinel-1, with seven-band imagery, multiple resolutions and polarization, and techniques to filter to a single band.

Visualize multi-band Landsat raster data by creating a false color composite with near infrared, red, and green bands; vegetation appears bright and water stays black, with shortwave infrared highlighting features.

Explore basic image filtering in Google Earth Engine by date and area using Sentinel data for Japan; compute median reducers, create rgb composites, and address cloud cover for effective visualization.

Filter optical Landsat or Sentinel data to select images with minimal cloud cover using a cloud cover threshold and surface reflectance, applying filter bounds and a median reducer for clarity.

Learn to apply image reducers on image collections in Google Earth Engine, filter to Cambodia 2018, and reduce over space and time with mean, median, minimum, maximum.

Learn how to filter satellite images to country or region boundaries and clip to exact borders, using Japan as an example, to avoid spillage and obtain precise boundary-aligned imagery.

Define regions of interest interactively in Google Earth Engine, convert drawings to feature collections, render geospatial layers, and visualize digital elevation model palettes from interactive polygons.

Explore three main optical data types in Google Earth Engine—MODIS, Landsat, and Sentinel—and their data catalogs, resolutions, and products for vegetation indices, fire, land cover, and change detection.

Understand the principles of optical remote sensing, including passive sensing and multispectral bands, and how near infrared, red, and green reflectance reveal vegetation properties for forest and land-use mapping.

Preprocess Landsat optical data by correcting radiometric and geometric distortions, apply atmospheric correction to derive top of the atmosphere reflectance, and discuss dark object subtraction.

Identify Landsat sensors, learn band designations, and select near infrared and red bands for vegetation analysis using Landsat 8 and legacy Landsat data from Earth Explorer.

learn how spectral sharpening blends the 15-meter panchromatic band with 30-meter Landsat color bands using the huge saturation value algorithm to produce sharper imagery and better classification.

Adjust raster resolution and cell size in Earth Engine by resampling 500-meter vegetation index data to 30 meters, projecting forest cover data to a common crs.

Explore how vegetation indices derived from sentinel-2 surface reflectance in Google Earth Engine reveal greenness, burning, and flooding through ndvi, ndwi, and the normalized burn ratio.

Explore computing common vegetation indices with Google Earth Engine using Sentinel surface reflectance data, applying normalized difference formulas across bands b3, b2, b12, filtering by Bolivia, and visualizing results.

Learn to derive vegetation indices from optical data and create masks for water and burn scars using thresholding, clipping to Bolivia, and visualization in Google Earth Engine.

Explore synthetic aperture radar concepts, including active data collection, wavelength, and polarization. See how Sentinel-1 data on Google Earth Engine enables vegetation and urban monitoring with single or dual polarization.

Explore Sentinel-1 radar data in Google Earth Engine, featuring four polarizations and multiple resolutions from the Copernicus Sentinel collections, with backscatter informing surface structure unaffected by cloud cover.

Learn how to read sentinel-1 data, filter for the desired polarization bands, assemble a mosaic to create a spatially continuous image, and visualize features like Tonle Sap Lake in Cambodia.

Apply Sentinel-1 radar data to map floods in Cambodia using focal median smoothing and morphological filtering to highlight flooded areas near Tonle Sap.

Learn to work with both VV and VH polarizations from Sentinel-1 data, apply filters, mosaic, clip to produce a continuous radar image, and compute a degradation index.

Learn to access ALOS PALSAR data (2007–2018) at 25 m, convert to backscatter for HH and HV, and map Cambodia’s water and vegetation using Google Earth Engine.

Explore how Google Earth Engine aggregates diverse datasets—Landsat, Sentinel, night lights, forest change, population, and malaria data—and map accessibility to cities using the 2015 Global Accessibility to Cities dataset.

Analyze nightlights in Japan by converting a 2012 stable lights raster to vectors, applying three thresholds (30, 55, 62), and computing mean nightlight values.

Map monthly burnt areas using the MODIS data from the Modise family, focusing on Cambodia, and compute mean statistics across an image collection to visualize burn scars and their progression.

Learn the basic theory of machine learning for spatial data analysis, including unsupervised clustering of spectral-band pixels and supervised classification and regression with training data.

Explore theory and applications of machine learning for remote sensing, compare random forests and support vector machines, and learn to build, train, validate, and apply learning models to raster stacks.

Explore unsupervised classification in Google Earth Engine by using clustering to identify land cover patterns in Amazonian Bolivia from a Landsat 2001 image without training data.

Understand supervised classification theory and its use of training sites and spectral signatures to assign pixels, and review common algorithms like minimum distance, maximum likelihood, spectral angle mapper, and EM.

Create training data on the fly in Google Earth Engine to build a supervised classification, converting points to feature collections and labeling forest, water, and degraded vegetation.

Prepare Landsat-based training data, create a 75/25 training-testing split, sample seven bands from the clipped amazonian Bolivia scene, and train a random forest classifier with 100 trees.

Explore classifying Landsat imagery in Google Earth Engine with a random forest model to map forests, water, sparse vegetation, and degraded vegetation, achieving about 92–95% overall accuracy.

Compare producer's accuracy and user accuracy to evaluate how well land cover classes map to ground truth and how reliable the resulting map is for users.

Learn how to export imagery, shapefiles, and feature collections from Google Earth Engine to Google Drive, manage export tasks, set scale and max pixels, and download training samples.

Discover spectral signatures and reflectance across wavelengths to distinguish vegetation and water, and learn to compute and visualize these profiles in a seven-band Landsat composite using Google Earth Engine.

Explore spectral mixture analysis to identify vegetation, bare earth, and water using Landsat seven band reflectance profiles and endmember values, and perform spectral unmixing to map land cover.

Introduce tensors as the generalization of vectors and matrices, showing them as arrays from vectors to 3d tensors and beyond, with gpu acceleration mentioned but not used in this course.

Learn to upload external geospatial data to Google Earth Engine, including shapefiles, manage assets and ingestion status, and share datasets for use in JavaScript or Python via Colab.

Explore object-based image analysis in Google Earth Engine, including Obeah concepts, object segmentation with snake clustering, and integrating spectral and spatial features to improve classification with machine learning.

Explore object-based image analysis in Google Earth Engine by preparing 2016 Ethiopian Landsat data, selecting bands, creating cloud-free composites, normalizing images, and forming clusters for object-based image analysis groundwork.

Compute distance raster from the Bangladesh coastline using an euclidean kernel, clip to the region, and visualize the resulting distance raster.

Explore how distributed computing frameworks coordinate parallel tasks, ensure fault tolerance, optimize resource allocation, and scale with growth through real-world examples like Hadoop and Spark.