You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The user will define the selection over the non-spatial dimensions; Lonboard/deck.gl-raster will define the per-chunk selection over the spatial dimensions.
When the user moves the frontend map, deck.gl issues a tile request, which reaches the Python side of Lonboard, and then Lonboard issues an async tile load for that tile. When the async task finishes, it sends the data back to the frontend to be visualized.
So Lonboard will issue a separate xarray async load for each of the spatial chunks that are currently visible in the viewport.
Because the user's Python session is fetching all of the chunks in the viewport concurrently, the callback must be async. Hence why I'm looking to use xarray's async API (ref Implement Xarray async integration #29)
Medium term work will move an entire 4D stack of raw pixel data to the browser, which can then be animated in the browser
I think this will be really cool to do arbitrary band math on animated COG data cubes on the fly
developmentseed/lonboard#1179 is starting to prototype what a Lonboard-xarray integration looks like. See https://github.com/developmentseed/lonboard/blob/d9eae7c022d6813b15ed5d2687348926e2c78249/dev-docs/superpowers/specs/2026-04-23-lonboard-zarr-support-design.md for ideas brainstormed with claude.
I think this will be really cool to do arbitrary band math on animated COG data cubes on the fly