Catalog API PU calculation

1

Hello!

I am a bit confused with the calculation of processing unit cost for catalog api on sentinelhub. The Processing Unit calculation page gives two multiplication factors: one for area of the query and another for ‘time period in months’.

If I take sentinel-2 L2A data and query for a period of 12 months (01-01-2023 to 01-01-2024 for example) and the area of interest is less than 10000 sq kms, then the multiplication factor for area would be 0.01 and the factor for months would be 12. that would lead to a PU cost of 0.01 * 12 = 0.12, correct? But as per the consumed processing units for the request using request builder, pu cost was less that 0.12.

Could you please help me understand with an example on how to calculate the incurred pu cost?

Regards,
Shreyas

2

Your assumptions seems to be correct.
Can you provide an example of the Catalog API request you’ve done (best to copy-paste from Request Builder, just make sure to remove the authentication token) and share your findings on the consumed PUs?

3

I have attached the screenshot of PU consumption, which the dashboard shows to be 0.01

PU_cost
PU_cost744×747 24.5 KB
.

The request from request builder is as follows:
curl -X POST https://services.sentinel-hub.com/api/v1/catalog/1.0.0/search
-H ‘Content-Type: application/json’
-H 'Authorization: Bearer
-d ‘{
“collections”: [
“sentinel-2-l2a”
],
“datetime”: “2023-04-01T00:00:00Z/2024-04-01T23:59:59Z”,
“bbox”: [
12.44693,
41.870072,
12.541001,
41.917096
],
“limit”: 100
}’

4

Any update on how the catalog pu is computed?

5

You have indeed identified a bug in our calculation. We are now working to fix it, thanks.
In general, your assumption is correct.

1 Like
6

Oh, I see. Glad I pointed it out.

As per the PU cost logic for Catalog API, does it matter which data is being requested? Since it only takes into account, the requested area and the requested time period, it doesn’t matter if it is sentinel (once in 5 days) , or planet (daily), right?

7

Yes, this is indeed correct.

1 Like