How Iceberg, GeoParquet & CARTO are reshaping geospatial
For years, geospatial data lived on the margins of the data ecosystem—locked in proprietary formats, reliant on specialized tooling, and disconnected from the cloud-native revolution reshaping analytics.
But the landscape is shifting. With the arrival of native geospatial support in Apache Iceberg and GeoParquet - and with CARTO playing a key role - spatial data is stepping into the future. Geospatial data types are now natively supported in Apache Iceberg and Parquet, marking a major turning point in how spatial analytics can be built and scaled. And at CARTO, we’ve been helping to push this shift forward from the start.
Here's how it's happening, and what it means for your data strategy.
If you’ve been working in geospatial for a while, you’ve probably heard the phrase “spatial is special.” It means that geospatial is - well - different.
While modern data teams embraced cloud-native technologies - from open file formats like Apache Parquet to table formats like Apache Iceberg - spatial data often lagged behind. It required proprietary formats, special tooling, and cumbersome data workflows. The benefits of scalable, decoupled architecture just weren’t available to teams working with spatial data - but that’s finally changing, in a big way. While the broader data community evolved toward open standards and modular architectures, geospatial was left out of the conversation.
That exclusion wasn’t just inconvenient - it was expensive and limiting. Geospatial teams were left struggling with:
- Duplicate datasets across multiple systems to support different workloads
- Paying for monolithic spatial warehouses that couldn’t scale flexibly
- Struggling with governance across inconsistent tools and formats
- Accepting vendor lock-in as the cost of doing location-based analysis
But now, thanks to community-driven efforts - including CARTO’s active role in the development of GeoParquet and Iceberg’s GEO spec - spatial data can finally join the modern data ecosystem.
Together, GeoParquet and Iceberg offer a modular, cloud-native architecture for spatial analytics.
- GeoParquet brings standardized, compressed storage for geometry and geography types.
- Apache Iceberg adds full table management, time travel, partitioning, and multi-engine access - all without vendor lock-in.
That means you can store your spatial data once, access it from anywhere, and avoid costly rewrites or duplication. Whether you're using Snowflake, Spark, DuckDB, BigQuery, or something else entirely, your data stays consistent, optimized, and open.
This shift isn’t just about cleaner architecture - it directly improves outcomes for teams working with spatial data:
- Lower storage & compute costs via efficient compression and query pruning
- Better governance through versioning, schema evolution, and unified access control
- Faster insights by enabling different engines to analyze the same data without conversion
In short, spatial data can finally behave like the rest of your data: accessible, governed, and scalable.
CARTO has been a key contributor to both GeoParquet and Iceberg’s geospatial evolution. Our Founder & CSO Javier de la Torre, who serves as chairman of the Open Geospatial Consortium Standards Working Group for GeoParquet, has been instrumental in connecting major data processing engines to adopt GeoParquet and facilitating the movement toward more open and interoperable geospatial data standards.
We’ve helped engines like BigQuery, DuckDB, and Snowflake to "speak" GeoParquet fluently, and have championed Iceberg’s new geospatial capabilities from day one.
The momentum is accelerating. Next week, Snowflake will officially announce support for GEOGRAPHY and GEOMETRY types on Apache Iceberg at the Snowflake Summit. The following week, Databricks will do the same at the Data + AI Summit. Google is also actively working toward supporting native geospatial types on Iceberg.
CARTO has collaborated closely with all of them - shaping the specifications, implementation patterns, and test cases that are bringing geospatial into the cloud-native mainstream.
But our mission goes beyond vector data. We’re helping to standardize the next wave of geospatial formats:
- Raster: through initiatives like Parquet Raster, we’re bringing imagery and gridded datasets into scalable, columnar storage.
- Mobility & Trajectories: CARTO is pioneering support for trajectory data - enabling use cases like telemetry ingestion, movement analysis, and visualization through our new Workflow Extensions.
- 3D & Lidar: We're exploring how spatial indexes, point clouds, and digital twin data can be made interoperable and queryable in the same architecture
With this announcement, vector data becomes essentially a first-class data citizen on the cloud. The next challenge? Raster. CARTO is already working with partners to extend the benefits of Parquet to imagery and gridded datasets - you can learn all about this initiative here.
Embracing Apache Iceberg and GeoParquet represents a transformative moment for spatial data analytics. No longer sidelined, geospatial data can now fully participate in the modern, cloud-native data ecosystem - unlocking new levels of efficiency, governance, and interoperability.
Want to learn what this shift could mean for your organization? Request a demo from one of our geospatial experts!
