Spade is ideal for remote sensing workflows that need to process large volumes of satellite imagery efficiently.
The Challenge
Remote sensing pipelines typically involve:
- Downloading imagery from multiple satellite providers
- Reprojecting data into consistent coordinate systems
- Tiling large rasters into manageable chunks
- Running analysis algorithms across thousands of tiles
- Combining results into final products
These workflows are inherently parallel but traditionally require complex orchestration code.
How Spade Helps
With Spade, you define each processing step as an independent block and wire them together in a YAML pipeline. The scheduler automatically:
- Parallelizes tile processing across all available workers
- Caches intermediate results so re-runs skip completed steps
- Handles failures gracefully with per-block error isolation
- Scales from a laptop to a cluster without pipeline changes
Example Pipeline
A typical remote sensing pipeline in Spade might look like:
- Fetch — Download imagery using a data provider block
- Reproject — Transform to the target CRS using the GDAL reproject block
- Tile — Split into a collection of tiles using a map block
- Analyze — Process each tile in parallel
- Merge — Reduce results back into a single output