There are many tools available for rendering geospatial data in web maps, but scientific use cases like mapping multi-dimensional climate data push existing tools to their limits. Our new [open-source library](https://github.com/carbonplan/maps) supports loading numerical, multi-dimensional raster data stored in the Zarr format and directly rendering data tiles in the browser using WebGL. This approach supports rendering data with arbitrary dimensions and provides a lightweight interface to configure the rendering of that data, including the ability to perform calculations to combine data across dimensions. In this talk, we present an explanation of how this library integrates with Zarr data, some sample use cases, and next steps for the library. Live slides and demos can be viewed here.