SPAI AS A FRAMEWORK FOR DEVELOPING APPLICATIONS BASED ON SATELLITE DATA AND ARTIFICIAL INTELLIGENCE
Fran Martín Rivas
Product & Data
fran@earthpulse.ai
Turning EO and AI into actionable solutions across industries
ABOUT ME
Product Manager and Data Scientist at EarthPulse, with the mission of bridging the gap between Earth Observation data and user needs.
Fran Martín Rivas
ABOUT US
We are the result of a fusion of diverse backgrounds and over 20 years of experience in Earth Observation, Artificial Intelligence, business, and innovation, united by a mission to democratize satellite analytics — making them accessible, useful, and easy for users to understand.
EarthPulse
THE PROBLEM SPACE
… let’s start with a story …
A couple of years ago, we were involved in an operational service for an insurance company. They managed hundreds of thousands (even millions) of assets distributed around the world and wanted to know what climate risks they were exposed to. Moreover, they wanted to be able to consume this via a dashboard and also be able to have it updated in their database!
MULTI-RISK CLIMATE ASSESSMENT
This supposed several challenges...
A couple of years ago, we were involved in an operational service for an insurance company. They managed hundreds of thousands (even millions) of assets distributed around the world and wanted to know what climate risks they were exposed to. Moreover, they wanted to be able to consume this via a dashboard and also be able to have it updated in their database!
MULTI-RISK CLIMATE ASSESSMENT
1
Specific knowledge for each climate risk
2
Varied and dispersed sources and types of data
3
Need to deploy very different services: dashboards, APIs...
4
Lack of computing and storage capacity at scale
It was certainly difficult!
There are markets affected by territory dynamics, and they need to protect their assets and minimize economic impact.
There is a need of geoinformation products to understand location risk and territory changes
EO data is an untapped opportunity, but remains with a low adoption level due to the lack of knowledge, about which is its value and how to get it.
The process is broken at this point
OUR USERS POINT OF VIEW
There is a low adoption of geoinformation products based on Earth Observation data …
OUR POINT OF VIEW
… because it is hard to develop and deliver them at scale, not to mention understanding user needs and connecting them with these data
Developing and deploying AI powered Earth Observation applications can be challenging and resource-intensive.
Data access, handling diverse data formats, processing with high computational demands and integrating AI models is a complex task.
There is a pressing need for a comprehensive, user- friendly framework that streamlines these processes.
DATA FRAGMENTATION
RESOURCE INTENSIVE
LACK OF DEVELOPMENT FRAMEWORK
SPAI
… and why it was born …
Cloud-native & cloud-agnostic
Auto-scalable resources
On-premises deployment
Reliable, Robust
ACCESS,
PROCESSING and VISUALIZATION of EO data.
Leverage Geospatial Intelligence
Easy, fast and custom solutions
Data Scientists Developers
EO Experts
SPAI SDK
Have data knowledge? Work more efficiently
Use your favourite code editor
Just select what, where, and when. We make it happen
SPAI HUB
1
Choose your
ready-to-use solution
2
Define where and when
DEPLOY
3
Start making informed decisions
SPAI CHAT
Ask anything to SPAI CHAT
Data Discovery
EO Expertise
SPAI Technical Support
API
S3
PLUGIN
Integrate the results wherever you are
currently working
REAL-WORLD USE CASES
LANDSLIDE RISK EARLY WARNING SYSTEM
Leading global operator in the management of road infrastructures. The company oversees more than 8,000 kilometers of roads worldwide.
Ensuring the stability of slopes along its road networks it’s a critical responsibility. The company must constantly monitor, assess, and manage the susceptibility of slopes to landslides to prevent accidents and protect both travelers and nearby communities.
THE CLIENT
THE NEED
LANDSLIDE RISK EARLY WARNING SYSTEM
The objectives were to develop and validate satellite technology as a key tool for assessing slope stability and detecting risks, implementing an early warning system that continuously and predictively calculates the probability of a landslide.
To achieve this, we have trained and implemented an ML model capable of calculating landslide risk based on a multitude of geospatial and satellite data.
LANDSLIDE RISK EARLY WARNING SYSTEM
How is SPAI used?
1
Unified data access: Sentinel-1, Sentinel-2, Copernicus DEM, ERA5 …
2
Cloud resources management and computational demands handling for data storage, AI model traning, processing workflows and inference
3
Service deployment, with dashboards, APIs and alerts
4
Full integration with SPAI Chat, our NLP data interface
LANDSLIDE RISK EARLY WARNING SYSTEM
From this …
reactive
LANDSLIDE RISK EARLY WARNING SYSTEM
… to digitalization, with a complete Early Warning System
proactive
LANDSLIDE RISK EARLY WARNING SYSTEM
Interactive dashboard where you can see intermediate indicators, such as ground motion (inSAR), vegetation health or humidity
LANDSLIDE RISK EARLY WARNING SYSTEM
Interactive dashboard where users can see intermediate indicators, such as ground motion (inSAR), vegetation health or humidity …
LANDSLIDE RISK EARLY WARNING SYSTEM
… alongside an aggregated and dynamic landslide risk index, where alerts are triggered depending on the forecasted risk.
LANDSLIDE RISK EARLY WARNING SYSTEM
Fully integrated with SPAI Chat, our latest leap forward in usability and explainability
LANDSLIDE RISK EARLY WARNING SYSTEM
What is the impact?
1
Digitalization of slope risk management
2
From reactive to proactive operations
3
Operational efficiency and resource optimization
4
A tangible return of investment, delivering up to 9 times the value of the ops cost
5
Reduction in economic losses due to landslides of 70%
Global leader in digital water cycle management, providing advanced technological solutions to utilities and water companies across more than 30 countries.
Despite the advanced digital tools already integrated into user’s operations, identifying and managing leaks in large-scale water distribution networks remains a major challenge. Leaks often go undetected for long periods, leading to significant water loss, increased operational costs, and environmental concerns.
THE CLIENT
THE NEED
PIPELINE WATER LEAK RISK MONITORING
The objectives was the development of algorithms and training of an AI model to calculate network vulnerability and the risk of water leaks in the pipeline network, ultimately resulting in a system capable of creating risk maps and identifying areas with a high risk of water leaks.
To achieve this, we have trained and implemented a DL model capable of generate a vulnerability and leak risk map to identify the most vulnerable areas of the pipeline network with a high risk of water leakage.
In operations, this model can be executed anywhere our user has customers.
PIPELINE WATER LEAK RISK MONITORING
How SPAI is used?
1
Unified data access: Sentinel-1, Sentinel-2, Copernicus DEM, ERA5, ESA Worldcover, soil data …
2
Cloud resources management and computational demands handling for data storage, AI model traning, processing workflows and inference
3
Replication of the use case at scale anywhere our user wants to get data from
4
PIPELINE WATER LEAK RISK MONITORING
Seamless integration within user’s digital product
The service can be accessed and executed in different ways, depending on the technician's needs, and can be performed at any time. Therefore, the technician must establish an area of interest and a timeframe to deploy and execute the service.
PIPELINE WATER LEAK RISK MONITORING
PIPELINE WATER LEAK RISK MONITORING
Improved operational efficiency. Instead of covering the entire city or area of interest, the field team knows which areas to prioritize.
Pipeline network categorized by leak risk. This is just a sample!
Which is the impact?
1
Improve network vulnerability assessment
2
Improve operational efficiency
3
4
PIPELINE WATER LEAK RISK MONITORING
Reduction in water losses of +20%
Expand the range and offering of services, creating new business lines
WHAT’S UNDER THE HOOD?
… so you don’t have to worry …
1
Unified data access
2
Cloud-based infrastructure
3
AI Model training, recurrent processing,
app hosting
4
In and out integrations
All in one place!
KEY TECHNICAL INSIGHTS
Whether federating third-party solutions or integrating directly into user workflows, SPAI is frictionless first.
Computing, storage, service deployment, cloud-native geospatial formats… All scalable from our cloud infrastructure
Copernicus Data Space Ecosystem, Copernicus Services, Planetary Computer, OpenStreetMap … whatever you need, with SPAI
THE IMPACT
FOR US
FOR OUR USERS
All the mentioned features are enablers! User needs are at the heart of everything we do
Reduces operating speeds by a 40%
Reduces monitoring costs by a 70%
Return of investment of 9:1
Allows to replicate the solutions in any region, anytime.
Allows to integrate the solutions withing users tools & workflows.
Allows to personalize the solutions to match user needs.
Reduce solutions development time by up to 60%
Decrease operational costs related to deployment and maintenance of solutions by up to 40%
Increase decision-making effectiveness based on EO analytics
Increase in the number of large-scale AI applications with EO data
Satellite analytics democratization
e
JOIN THE MOVEMENT!
Join our Discord channel!
Contact us! earthpulse.ai/
Created by users, for users. We value your feedback — collaboration and knowledge sharing are what power us!
Follow us on our social media!
AS A FINAL MESSAGE …
FIRST QUESTION?
BONUS TRACK: MULTI-RISK ASSESSMENT
… we achieved it with SPAI!
WWW.EARTHPULSE.AI
ALL CAPABILITIES OF SATELLITE DATA NOW AT YOUR FINGERTIPS
Fran Martín Rivas
Product & Data
fran@earthpulse.ai