AI Prompt Engineering for Google Earth Engine (GEE): Remote Sensing & Geospatial Analytics equips geospatial professionals, GIS analysts, remote sensing specialists, and researchers with practical prompt engineering techniques to accelerate satellite imagery analysis, automate geospatial workflows, and generate actionable insights using AI and Google Earth Engine.

 

Duration 3 days – 21 hrs 

 

Overview

This training equips participants with practical prompt engineering skills to use AI tools (e.g., ChatGPT-class assistants) for faster, more accurate work in Google Earth Engine—from writing and debugging GEE scripts to designing repeatable geospatial workflows for remote sensing analytics. Participants will learn how to translate geospatial problems into clear AI prompts, validate AI outputs, and integrate AI-assisted scripting into real-world GEE tasks such as land cover mapping, change detection, NDVI time series, water/flood monitoring, and export-ready outputs.

 

Objectives

• Apply core prompt engineering frameworks (role, context, constraints, examples, output format) for geospatial tasks.
• Use AI to generate, refactor, and debug Google Earth Engine (JavaScript / optional Python) scripts safely and efficiently.
• Design prompts for common GEE workflows: data ingestion, preprocessing, indices, classification, time-series, and exports.
• Validate AI-generated code using geospatial reasoning, QA checks, and reproducible documentation.
• Build reusable GEE components (functions, modules, parameterized scripts) with AI-assisted guidance.
• Produce a working mini-project (dashboard/map outputs or analysis notebook) aligned to their use case.

 

Target Audience

• GIS Analysts, Remote Sensing Analysts, Geospatial Data Analysts
• Environmental Scientists, Climate/Disaster Risk Teams
• Urban/Regional Planners, Agriculture/Forestry teams
• Data Analysts/Scientists working with spatial data
• Government/Academe researchers using satellite imagery and geospatial monitoring
• Developers supporting geospatial applications using GEE

 

Prerequisites 

• Basic GIS concepts (raster vs vector, projections, bands, resolution)

• Comfortable using a web browser and cloud tools

• Basic JavaScript or Python familiarity

• Basic remote sensing concepts (spectral indices like NDVI helpful)

• Prior exposure to Google Earth Engine interface

 

Course Outline 

 

Day 1 — Foundations: Prompt Engineering + GEE Setup & Core Workflow

 

Module 1: AI Prompt Engineering Essentials (for technical work)

• What AI can/can’t do for geospatial coding
• Prompt anatomy: role, goal, constraints, context, examples
• Output control: structured formats (steps, code blocks, checklists)
• Iterative prompting: refine → test → diagnose → improve
• Prompt patterns for code: “generate”, “explain”, “debug”, “optimize”, “convert”, “document”

 

Module 2: Responsible Use & Validation

• Avoiding hallucinations in datasets, sensors, and parameters
• Verification checklist: dataset IDs, band names, date ranges, scale, CRS
• Reproducibility: documenting assumptions and parameters
• Data governance basics (privacy, licensing, ethical geospatial use)

 

Module 3: Google Earth Engine Fundamentals (hands-on)

• GEE concepts: Image, ImageCollection, Feature, FeatureCollection
• Filtering by date, bounds, metadata; clipping and masking
• Visualization basics, map layers, palettes
• AI-assisted coding practice: “Write a script that loads Sentinel-2, filters clouds, and computes NDVI”

Lab 1 (Guided):

• Setup + first workflow: AOI → dataset selection → preprocessing → index → map visualization

 

Day 2 — AI-Assisted Remote Sensing Analytics in GEE

 

Module 4: Prompting for Dataset Selection & Preprocessing

• Choosing sensors (Sentinel-2, Landsat, MODIS) based on resolution/needs
• Cloud masking strategies and common pitfalls
• Compositing (median, mosaic), scaling, unit checks

 

Module 5: Indices & Derivatives (hands-on)

• NDVI, NDBI, NDWI, EVI (when/why)
• Zonal statistics for administrative boundaries / farms / watersheds
• AI prompts for reusable functions: “Create a function computeIndex(image, indexName)”

 

Module 6: Time Series & Change Detection

• Building time-series charts (monthly NDVI, seasonal composites)
• Simple change detection methods (before/after, trend)
• Exporting tables and rasters (Drive/Cloud) with correct scale and region

Lab 2 (Guided):

• NDVI time series for an AOI + summary stats per polygon + export outputs

 

Day 3 — Classification, Automation Patterns, and Capstone

 

Module 7: AI-Prompted Land Cover Classification

• Supervised vs unsupervised overview (practical focus)
• Training data creation in GEE; sampling strategies
• Classifiers (Random Forest basics), accuracy assessment (confusion matrix)
• Prompting AI for classification pipeline + troubleshooting

 

Module 8: Workflow Packaging & Documentation

• Turning scripts into reusable templates (parameters, functions)
• Prompting for refactoring: readability, modularity, performance
• Building a mini “analysis recipe” (prompt + code + validation checklist)

 

Capstone Mini-Project (choose one)

• Land cover map for AOI + accuracy report
• Flood/water extent monitoring using NDWI
• Vegetation health monitoring (NDVI trend) + zonal summaries
• Urban expansion indicator (NDBI over time)
  Deliverables:
• GEE script (clean + commented)
• 1-page workflow summary (inputs, steps, outputs, validation checks)