Hello, good afternoon!
I wanted to ask about the possibilities of processing areas larger than 2500 pixels in Sentinel Hub.
Hello, good afternoon!
For processing areas larger than 2500 px in Sentinel Hub, I can suggest 2 options (with a bonus alternative at the end ):
Large Area utilities
If your workflow is in Python and your area of interest (AOI) is not too large, you can make use of the sentinelhub-py package, and in particular the large area utilities functionalities. This tool allows you to split your AOI into smaller areas automatically (with some level of control on your side) and run separate Process API requests. I would say this option is practical for areas slightly larger than 2500*2500 pixels, but is inefficient for really large areas.
Batch Processing API
The Batch processing API was specifically designed to operate over large areas (all the way up to processing the entire World). Although this API requires an Enterprise account and your own object storage (e.g. CreoDIAS or AWS S3), it is incredibly efficient in processing large AOIs in very little time. The service automatically tiles your area into a grid (of your choosing) and after processing in parallel in the cloud, delivers the results to your bucket. You can read more about an example of our service in our blog.
Bonus: Asynchronous Process API (in BETA)
We are developing an asynchronous equivalent to the Process API, intended for larger requests where you don’t need the results immediately. This service was announced in our last two newsletters in June and September. This service, still in Beta, works similarly to the Process API, with the difference that limits on resolution, output raster dimensions, and time interval of input data are relaxed. The service delivers results to object storage, like Batch. If you are interested, please drop us an email.
Hope this helps,
Thanks for your answer!
What is the difference between Batch Processing API and Asynchronous Process API?
You can consider Asynchronous Process API as an intermediate tool between Process API and Batch Processing API: it was designed for cases that are too large for Process API (more than 2500 pixel limit) but not large enough to justify using Batch Processing.
Currently the tool is limited to 10 000*10 000 pixels2 (4 times Process API) and does not tile the results like Batch Processing API, but delivers a single product to your object storage.
You can see the functionalities of the tool in the API reference documentation. As mentioned before, don’t hesitate to ask if you want to test it out!
Thank you very much for the reply!
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.