--- title: "PAINLESS GEOSPATIAL ANALYTICS USING MOTHERDUCK’S NATIVE INTEGRATION WITH GALILEO.WORLD" canonical: "https://motherduck.com/blog/galileo-world-geospatial/" related: - title: "Getting started with modern GIS using DuckDB" url: "https://motherduck.com/blog/getting-started-gis-duckdb/" - title: "Google Gemini + MotherDuck Integration | DuckDB Analytics" url: "https://motherduck.com/ecosystem/gemini/" - title: "Interactive Data Visualization: 18M Rows in the Browser | MotherDuck" url: "https://motherduck.com/case-studies/dominik-moritz/" gated_asset: title: "DuckLake on MotherDuck" url: "https://motherduck.com/product/ducklake/" --- # PAINLESS GEOSPATIAL ANALYTICS USING MOTHERDUCK’S NATIVE INTEGRATION WITH GALILEO.WORLD From urban planning to climate analysis, real estate analytics to logistics, site selection to advertising — geospatial data is everywhere. But working with it has traditionally been hard: * Regular BI tools lack extensive geospatial capabilities * Geographic information systems (GIS) usually have a steep learning curve * Transformation issues between various formats * Poor performance with big datasets Whether you're a developer building spatial analytics or a business user exploring location-based trends, it's often a struggle when you need to get and share insights out of a geospatial dataset. ## Galileo.world – GIS meets DuckDB Traditionally, geospatial analysis meant spinning up a dedicated infrastructure: PostGIS databases, servers and scripts for data conversion. With DuckDB spatial extension, your device alone becomes a [powerful spatial tool](https://motherduck.com/blog/geospatial-for-beginner-duckdb-spatial-motherduck/). Galileo.world takes advantage of [DuckDB-Wasm’s](https://duckdb.org/docs/stable/clients/wasm/overview.html) capabilities of running queries directly in the browser and MotherDuck’s infrastructure to leverage performance for bigger datasets. Its technology is mostly based on these foundations: * DuckDB-Wasm: In-browser analytics engine for fast, serverless queries * MotherDuck: Native integration for scale * Deck.gl: GPU-accelerated layers for smooth, large maps Therefore, most of the action occurs in your browser, which results not only in performance, but also privacy, since files and maps do not leave it, unless you decide to share them. How regular GIS works: ![image3.png](https://motherduck-com-web-prod.s3.amazonaws.com/assets/img/image3_8d45e3f133.png) How galileo.world works: ![image2.png](https://motherduck-com-web-prod.s3.amazonaws.com/assets/img/image2_8a637f3d94.png) Galileo.world’s key features: * **Private by design**: Everything runs in your browser — no data leaves unless you share. * **Simple file input**: Load Parquet, GeoJSON, CSV, KML, SHP — directly in the browser * **MotherDuck native**: Hassle free geospatial analytics with your MotherDuck datasets. * **Custom visualizations and analytics**: Create responsive maps, charts and dashboards from geospatial data * **Simple sharing**: Share public projects or keep them local * **Public data catalog**: Add layers from a growing public data catalog to your projects ![image4.png](https://motherduck-com-web-prod.s3.amazonaws.com/assets/img/image4_545c5e538c.png) ## Working with big geospatial datasets – the pain points When working with geospatial data, two things kill performance: **high amount of** and **high complexity of geometries**. It’s common to see the following issues related to them: * Plotting everything causes memory bloat and UI stops responding * Maps get excessively slow when zooming or panning * Geometries overlap, creating more confusion than understanding In practice, **raw plotting** of big datasets creates **significant bottlenecks for real-time interactivity,** turning exploration and analysis into a struggle. The most common strategy for this case scenario is create **tiles**. A **tile** is simply a **small piece of a bigger dataset**, divided by predefined grids at each zoom level. Each tile contains a limited number of geometries and edges, usually defined when you create it. That limitation allows tiles to render faster while still visually convincing for bigger datasets. Even though tiles work very well for visualization, they are not designed **for** **analytical purposes**, since they do not necessarily contain all the data from the original dataset. Therefore, performing calculations over tiles can provide misleading results due to incomplete data. A more comprehensive guide to tiling can be [found here](https://carto.com/blog/map-tiles-guide). ## Visualization + analytics for all sizes of geospatial data – the dual execution engine In order to display big datasets and still maintain analytical fidelity to the original data, galileo.world adopts a **dual execution engine**. Taking advantage of DuckDB-Wasm and MotherDuck full capabilities, the app operates with multiple workers, orchestrating queries that’ll plot geometries on the map and those that will provide analytical outputs such as charts. For visualization, the dataset goes through **sampling** and **geometry simplification**, which virtually eliminates any dataset size limitations and increases performance while dynamically zooming or panning. For analytics, not only the data displayed on the map is used, but the entire **original** **dataset**, hence preventing misleading calculations and missing data. ![image1.png](https://motherduck-com-web-prod.s3.amazonaws.com/assets/img/image1_4bfe143228.png) Whether working with big or small geospatial data, the combination of MotherDuck and galileo.world is a powerful duo to make your data analysis, visualization and project sharing faster, simpler and more secure. [Try it here](https://galileo.world) to see what’s possible and [join galileo.world’s slack community](https://join.slack.com/t/galileoworldcommunity/shared_invite/zt-3bb1geymp-_92RGgohgyxNxItxv3J0dQ).