Google launches Custom Satellite Embeddings powered by AlphaEarth Foundations

Google has launched the Private Preview of Custom Satellite Embeddings, allowing organisations to generate on-demand geospatial insights using the AlphaEarth Foundations AI model for monitoring and change detection.

Google has announced the Private Preview of Custom Satellite Embeddings, a new product in its geospatial data offerings that leverages its AlphaEarth Foundations AI model. The offering is intended to enable organisations to better track how the environment and geography of a region change over time by providing them with a summary of the satellite data that can be accessed when they need it. Also Read: Google Search’s AI Mode To Show Ads Soon? All You Need To Know

The development is an extension of the annual Satellite Embeddings dataset launched last year via Google Earth Engine and Google Cloud Storage, and will now allow for ongoing monitoring and change detection.

On-demand satellite monitoring by Satellite Embeddings

The new Custom Satellite Embeddings feature enables businesses to request an embedding of satellite data for specific timeframes, such as quarterly, monthly, weekly or even five day intervals, once a month, depending on the input data available.

Google's flexibility allows businesses and public sector organisations to shift from the traditional one-off mapping to monitoring dynamic landscape changes.

The new product complements its global dataset with higher frequency insights that provide users with more granular information to suit their areas of interest, the company said.

AlphaEarth Foundations combines multiple satellite data sources

The new product is founded on Google's AI model, AlphaEarth Foundations, which integrates various categories of satellite data into a single model.

The model incorporates optical imagery, radar and elevation data and 3D laser mapping (LiDAR) to create a 64-band geospatial image. AlphaEarth Foundations combines these data sources to offer an integrated, coherent and complete representation of the Earth's surface, including areas with frequent cloud cover that make traditional satellite-based data difficult to use.

The AI model reduces the need for complicated data preprocessing and simplifies remote sensing workflow for organisations, according to Google.

Organisations can monitor crops, forests and disaster affected areas

Google pointed to a number of use cases for Custom Satellite Embeddings, including in the industry, the education sector, and, of course, the advertising world.

For agriculture, organisations can use crop classification, near real-time growing season data and damage estimation from frost and storms. The technology can also be used for agricultural finance and market analysis, offering real-time insights throughout the agricultural cycle.

It can be used by public sector agencies for forest health monitoring, urban development and assessing damage after natural disasters.

The embeddings can be used to determine climate-related risks in consumer goods and retail supply chains, track agricultural-related disruptions, and enable compliance with the European Union Deforestation Regulation (EUTR).

Utility and telecommunications companies can track vegetation growth and the changes of water bodies and construction activity that could impact infrastructure.

Satellite Embeddings have already been used for a number of use-cases

Google has already cited that organisations have been testing the annual Satellite Embeddings dataset to enhance a variety of geospatial analysis applications.

The company says that they have used the technology to enable partners to map forest carbon, identify landscape change and identify agricultural facilities without the need for a specialised machine learning expert or large computing resources.

The AI generated embeddings can take complex data from a satellite and transform it into something useful for analysis by organisations.

Enterprises and researchers can now access Private Preview

Google has announced the registration opening for organisations that want to join the Enterprise Private Preview of Custom Satellite Embeddings.

The firm also launching a dedicated research application programme. Selected researchers will have access to a free sample dataset via Google Earth Engine for their research.

Candidates are encouraged to have prior experience with Earth Engine, be a member of a university or research institution, and plan to publish or share their research results. The application for the academic programme will continue to be accepted until 1st September 2026.

Google adds AI-driven geospatial features to its tools

Google is now introducing Custom Satellite Embeddings, which will allow organizations to access satellite data more often and tailored to their needs, adding to the already vast annual mapping capability of AlphaEarth Foundations.

The firm believes the new entity will contribute to the efficient management of environmental, infrastructure and natural resources by companies, governments and researchers, with the added advantage of making analysing satellite data easier.

Get latest Tech and Auto news from Techlusive on our WhatsApp Channel, Facebook, X (Twitter), Instagram and YouTube.