Automated Land Classification: Open Earth Foundation | Artificial Intelligence Cloud Innovation Center

Overview

The Arizona State University Artificial Intelligence Cloud Innovation Center, powered by Amazon Web Services (AWS), collaborated with the Open Earth Foundation, a California-based nonprofit committed to open-source digital systems for planetary resilience. The goal of this proof of concept (POC) was to develop an AI-powered solution that uses satellite imagery and boundary data to classify land types and calculate land usage—supporting Open Earth’s work in greenhouse gas inventory modeling and sustainability reporting.

Problem

Open Earth Foundation lacks a scalable, automated technology platform to process satellite imagery for land usage data. Generating accurate land classification within user-defined boundaries is essential to support environmental policy, emissions tracking, and resource management—but the manual process is time-consuming and inconsistent. They needed a tool that could process geographical inputs, analyze satellite imagery, and produce standardized outputs in real time.

Student Spotlight

The AI CIC is powered by ASU Student Workers. The following students were assigned to this project to develop this open-source solution in partnership with the AWS and ASU mentor team.

Role: Backend Developer / Researcher | Connect with Bala on LinkedIn

I led research and development for advanced land classification, evaluating multiple methodologies before architecting an efficient pipeline with 10m pixel-level analysis from Sentinel data using the Dynamic World dataset. Harnessing Google Earth Engine's power, I engineered satellite imagery processing for accurate land type categorization, implementing scalable backend logic on AWS Lambda. As the core researcher and architect, I resolved complex integration challenges, facilitating smooth data delivery to interactive frontend maps. This immersive experience sharpened my skills in geospatial tech, cloud systems, and end-to-end connectivity, transforming raw satellite data into dynamic, insightful visualizations that drove project innovation.

Role: UI/UX Designer / Frontend Developer | Connect with Shristi on LinkedIn

I had the opportunity to conduct user research that revealed key pain points around geospatial data uploads and visualizations. From there, I defined an interactive design system with color tokens, grid rules, and reusable components to ensure a cohesive experience. I then led rapid ideation sprints, wire framing and refining the workflow based on client feedback until the interface felt intuitive and polished. Next, I translated those designs into a fresh React codebase, building JSX components and orchestrating state hooks so the UI responded seamlessly. The true thrill came when I uploaded a GeoJSON file and watched the map render alongside a precise area-calculation JSON, proof that design and logic had finally come together! Managing Git commits as milestones and deploying the live build with a single AWS Amplify click, I genuinely enjoyed seeing our client’s vision take flight.

Backend Developer | Connect with Ridham Sonani on LinkedIn

I was responsible for implementing land classification into different types using Dynamic World and Google Earth Engine. I processed and analyzed satellite imagery through Google Earth Engine to obtain land classification data and images, effectively categorizing the land into various classes. I developed the backend logic and deployed it as an AWS Lambda function, ensuring it was scalable and efficient. One of the most exciting parts was seeing the land classification data appear visually on the frontend as a real-time, interactive map after integrating everything together, which showed me how my code powered the application. Throughout this project, I learned a lot about working with geospatial data, cloud deployment, and connecting backend services with a frontend application. Overcoming integration challenges taught me valuable troubleshooting skills and deepened my understanding of full-stack development, while also giving me hands-on experience in remote sensing and smooth data flow for end users.

Approach

The AI CIC team developed a web-based platform where users upload a GeoJSON file with boundary coordinates and select a date range. The system then processes satellite imagery using Google Earth Engine (Dynamic World and Sentinel-2 datasets) and AI/ML logic hosted on AWS Lambda. Outputs include a JSON file summarizing land types and area in square kilometers, and a visual map that highlights land types using color-coded classification.

Key AWS Services Used:

AWS Lambda: serverless compute for backend logic
Amazon S3: data storage for user inputs and outputs
AWS CDK: infrastructure as code
CloudWatch: logging and error tracking

Industry Impact

This POC directly supports sustainability efforts by enabling environmental organizations like Open Earth Foundation to:

Generate data for greenhouse gas inventory reports
Understand land transformation over time
Improve the speed and reliability of environmental modeling
It replaces manual, time-consuming mapping tasks with an automated and repeatable solution—freeing up resources for deeper environmental analysis and planning.

"We are deeply grateful to the Cloud Innovation Center team for their outstanding work and collaboration. This project was an incredible opportunity to co-create an open-source solution with the CIC students in such a complex sector as AFOLU (Agriculture, Forestry, and Other Land Use) for greenhouse gas emission inventories. The experience not only strengthened our technical capacity but also showed the power of innovation and partnership in advancing sustainability solutions."
Maureen Fonseca Mora, Remote Sensing Data Engineer, Open Earth Foundation.

Wider Application

The solution has broad potential beyond this initial engagement:

NGOs and academic institutions can use it to support research on deforestation, urban sprawl, or agricultural trends.
Governments and policy agencies can integrate it into climate action planning and land management strategies.
Environmental consultants can adopt it as a plug-and-play tool for field assessments and sustainability reporting.

Next Steps

This proof of concept represents the first step in building a comprehensive overview of land classification at the city or municipal level. Building on this foundation, our next efforts will focus on integrating complementary open-source datasets to analyze land-use changes over time, assess environmental impacts, and calculate GHG emissions.

By applying globally available open-source data and Tier 1 methodologies, we aim to reduce friction between data availability and mitigation planning. With this tool, we can advance our work in a more scalable way, providing cities with insights and emissions data aligned with IPCC and GPC methodologies, and ultimately supporting more effective climate action.

About the ASU CIC

The ASU Artificial Intelligence Cloud Innovation Center (AI CIC), powered by AWS is a no-cost design thinking and rapid prototyping shop dedicated to bridging the digital divide and driving innovation in the nonprofit, healthcare, education, and government sectors.

Our expert team harnesses Amazon’s pioneering approach to dive deep into high-priority pain points, meticulously define challenges, and craft strategic solutions. We collaborate with AWS solutions architects and talented student workers to develop tailored prototypes showcasing how advanced technology can tackle a wide range of operational and mission-related challenges.

Discover how we use technology to drive innovation. Visit our website at ASU AI CIC or contact us directly at [email protected].

Photos

On This Page: