What changed in September 2026

• Line layers have an opacity slider under Color, and an opacity of 0 is kept when you save.
• Switching the colour palette keeps your custom break values.
• Unsaved changes to a custom colour scale are kept when you switch No data on or off.
• When you choose a field to size by, the size range starts at the current size instead of at 0.
• The data table keeps the current page visible while the next one loads.
• Long link labels in a dashboard’s Links widget wrap instead of running past the panel.
• Where a name is cut off, the full name shows when you hover over it.

• The scale bar shows full numbers instead of abbreviations with K and M.
• In a workflow, a catchment area with several starting points from a previous step now fills in its input correctly.
• Catalog search gives more weight to datasets that cover the area you are looking at.

When you calculate separate catchment areas for several starting points, each area now has an origin field that says which starting point it comes from. You can label, filter or style the areas by it.

• Catalog text search finds datasets more reliably.
• The Join tool is faster and handles large datasets without running out of memory.
• Tools no longer add extra bounding-box columns to their results.

Until now, accessibility analyses ran on the routing networks we provide, and that also set the limits of where GOAT could be used. In v3 you upload your own network and the analyses run on that.

Take Fukushima. We do not provide a routing network there, so the accessibility indicators had nothing to run on. Upload the street network and a Catchment Area works the way it does in Munich. The same holds for public transport: bring the GTFS data for the region and the transit indicators run on it.

That covers two different situations. One is geographic, everywhere we do not already cover, which is most of the world. The other is local: a planned cycle network that does not exist yet, a campus network that is not in OpenStreetMap, or a national dataset a planning office trusts more than the open one. The network arrives in the Overture format, a .zip containing segments.geoparquet and connectors.geoparquet.

Opening GOAT now starts on a Home screen instead of dropping you straight into an empty map.

It searches the whole workspace from one box, and puts the three things you are most likely to want next in reach: a new project, a dataset, or the Catalog. Underneath, Jump back in lists the projects you had open most recently, and Start from a template offers the workflow, dashboard and layout templates.

Links to the getting-started guide, the documentation and a short isochrones video sit on the same screen, so the first hour with GOAT does not depend on knowing where to look.

An empty map is a bad place to start, especially in your first hour with a new tool. v3 adds templates at three levels.

A project template opens a prepared project: the layers are styled, the analyses have been run and the Dashboard is built, so you can see how results are meant to be read as well as how they were produced. A workflow template is a prepared analysis where the only thing missing is your data, so ÖV-Güteklassen for your region does not mean building the chain of steps yourself. A layout template is a print-ready map you do not have to design from scratch.

Workflows and Layouts can now be added from inside a project, rather than being separate places you go to.

Until v3 the Catalog held around 60 datasets, each one found, checked and added by hand. It now holds more than 30,000.

They are no longer added one at a time. GOAT collects from public geodata catalogs instead: national open data portals across Europe, regional and municipal portals, Overture Maps and OpenStreetMap. Everything lands in one index, so a dataset from a German state portal and a layer from Overture are described the same way and can be found the same way.

What changes is the first step of a project. Search the Catalog, filter by area, by topic and by data type, and add the dataset to your project. No download, no reprojection, no cleaning.

Layers, datasets and projects are organised differently now, and the point of the rebuild was to make moving through them easy. Drag and drop to move things, put folders inside folders, and select several items at once to move or delete them together.

More importantly, everything is in one view. Projects, Templates, Datasets, Bundles and Folders sit together and you filter by type to see only what you want. That makes one question much easier to answer than it used to be: which data is connected to which project.

Sharing works across all of it now, not only on projects, and ownership of an item can be transferred to someone else. Useful when a colleague leaves, or when a project moves from the person who built it to the team that runs it.