Google Earth Engine for Satellite Image Analysis Training Course

Course overview

Satellite imagery programmes produce frequent, high-volume observations, but many GIS teams still rely on manual downloads, desktop processing and one-off map outputs. Google Earth Engine (GEE) changes that model by providing a cloud platform for searching, filtering, processing and analysing multi-petabyte Earth observation archives without maintaining local image stores. This course enables analysts to turn Sentinel-1, Sentinel-2, Landsat and climate datasets into repeatable analyses for land cover monitoring, vegetation assessment, flood mapping, urban growth studies and environmental reporting.

Participants learn to write and structure Earth Engine JavaScript workflows in the Code Editor, work with Image and ImageCollection objects, apply spatial and temporal filters, calculate spectral indices, mask clouds, compose mosaics and extract zonal statistics. The course develops practical capability in supervised classification, change detection, time-series analysis, accuracy assessment and scalable export workflows. Participants also learn when to use client-side versus server-side operations, how to avoid common GEE performance errors, and how to organise scripts for peer review and operational reuse.

Delivery combines instructor-led demonstrations with guided coding labs based on realistic monitoring scenarios. Participants build, test and refine an end-to-end Earth Engine workflow using a defined area of interest and satellite dataset, with checkpoints for interpretation, validation and output quality. They leave with a documented GEE script repository, exported map and tabular outputs, and a practical workflow design they can adapt for a live organisational use case.

The course is designed for GIS, remote sensing, environmental and data professionals who already understand spatial data and want to move from desktop image processing to reproducible, cloud-based satellite analysis.

Course objectives

By the end of this course, participants will be able to:

• Construct Google Earth Engine JavaScript scripts that load, filter and visualise ImageCollection datasets
• Apply spatial, temporal and metadata filters to Sentinel, Landsat and climate data collections
• Calculate NDVI, NDWI, NBR and custom band-math indices for analytical mapping
• Implement cloud, shadow and quality masking using QA bands and Sentinel-2 cloud probability data
• Produce seasonal composites and mosaics using median, percentile and quality-based reducers
• Classify land cover with labelled training samples and a Random Forest model
• Evaluate classification outputs using validation samples, confusion matrices and accuracy metrics
• Deliver a reusable Earth Engine workflow with exported raster, vector and zonal-statistics outputs

Benefits of attending

Target competencies

Who should attend

• GIS Analysts — who need repeatable cloud-based workflows for analysing large satellite archives
• Remote Sensing Specialists — who want to scale image preprocessing, classification and change analysis beyond desktop tools
• Environmental Scientists — who monitor vegetation, water, forests, habitat or land-use change across large areas
• Earth Observation Analysts — who need to convert Sentinel and Landsat data into defensible operational indicators
• Geospatial Data Scientists — who integrate spatial analysis, machine learning and time-series data into decision products
• Natural Resource and Sustainability Managers — who need to commission, review or apply satellite-derived monitoring outputs

Requirements and prerequisites

Participants should be comfortable working with GIS layers, coordinate reference systems, attribute tables and basic raster concepts such as pixels, bands, spatial resolution and classification. Prior experience with QGIS, ArcGIS Pro or similar GIS software is expected. Participants should also be able to read and make small edits to simple JavaScript-like code, including variables, functions and conditional logic; prior Python or JavaScript experience is helpful but not essential. No previous Google Earth Engine account, cloud-computing experience, advanced statistics or machine-learning background is required. Pre-course setup guidance is provided for creating a Google account and requesting Earth Engine access.

Training methodology

The instructor introduces each analytical pattern in the Google Earth Engine Code Editor, then participants implement it against supplied areas of interest and satellite collections. Labs progress from image filtering and visualisation to cloud masking, index calculation, compositing, classification and export. Short case discussions examine choices such as selecting a baseline period, defining training classes and interpreting false-change signals. Participants compare outputs in QGIS, review each other’s script logic, and complete an end-of-course application plan identifying a priority monitoring question, datasets, validation approach and production outputs for their workplace.

Course outline

Day 1: Earth Engine foundations and image collections

• Google Earth Engine architecture, catalogues and processing model
• Code Editor interface, script structure and Console diagnostics
• Earth Engine JavaScript objects and lazy server-side evaluation
• Geometry, Feature and FeatureCollection creation from areas of interest
• Image and ImageCollection data models
• Searching the Data Catalog for Sentinel, Landsat and climate collections
• Spatial, date-range and metadata filtering with filter functions

Workshop: Build a parameterised script that retrieves and visualises a Sentinel-2 collection for a chosen study area and reporting period.

Day 2: Preprocessing and spectral analytics

• Reflectance bands, scale factors and pixel-quality information
• Cloud and shadow masking with Landsat QA_PIXEL bands
• Sentinel-2 masking with SCL and cloud probability collections
• Band selection, renaming and image-property management
• Median, mean, percentile and quality mosaic compositing
• NDVI, NDWI, NBR and bare-soil index calculation
• Reducer-based zonal statistics for administrative and management units

Workshop: Create a cloud-masked seasonal composite, derive vegetation and water indices, and export zonal statistics for a supplied polygon layer.

Day 3: Time series and change detection

• Temporal aggregation into monthly and seasonal image collections
• Charting region-reduced time series in the Earth Engine UI
• Baseline, comparison-period and anomaly design
• Image differencing and normalised change metrics
• Pre- and post-event analysis for fire, flood and vegetation loss
• Threshold selection and binary change-mask creation
• Connected-pixel filtering and minimum mapping-unit rules

Workshop: Produce a time-series chart and mapped change layer for a before-and-after land-surface event, including a documented threshold rationale.

Day 4: Land-cover classification and validation

• Class schema design and representative training-sample selection
• Digitising, importing and attributing training FeatureCollections
• Predictor-stack construction from spectral bands and indices
• Random Forest classifier training and parameter selection
• Classified-image styling and post-classification filtering
• Independent validation sampling and train-test separation
• Confusion matrices, producer accuracy, user accuracy and kappa interpretation

Workshop: Train and validate a Random Forest land-cover model, then produce a classified map with an accuracy-assessment table.

Day 5: Operational workflows, exports and delivery

• Server-side mapping, iteration and reducer patterns for scalable scripts
• Client-side versus server-side errors and performance troubleshooting
• Reusable functions, parameter blocks and script documentation
• Batch exports to Google Drive and Earth Engine Assets
• Raster export settings, projections, scale and pyramiding policy
• Vector and CSV export for zonal-statistics deliverables
• GitHub version control and STAC-aligned metadata for reproducible workflows

Workshop: Complete and present an end-to-end Earth Engine workflow that exports a map, statistics table and documented script package for a workplace-style monitoring question.

Tools & standards covered

Google Earth Engine Code Editor, QGIS, Google Colab, SpatioTemporal Asset Catalog (STAC)

A typical training day

Live online deliveries follow the same structure in the East Africa Time zone, with shorter screen blocks and longer breaks.

What the fee includes

How you can take this course

Certification

Participants who complete the full five days receive the Skillset Development Certificate of Completion, stating the course title, course code, dates and delivery format — suitable for professional-development records and employer reimbursement.

Frequently asked questions