Skip to main content

Hello everyone.
Using Python, I have to extract statistical data relating to very small areas (the crown of the trees) but I have to do it for many trees (about 10000 trees) and, for each tree, we need data for many days (from 2017-7-1 to today with P1D aggregation).

Through Request Builder I prepared the single API request (see below), I get the data but, since I have to make many requests for very long periods, I’d like to be able to reduce the amount of PU used for each request.

Can someone kindly help me understand where I can intervene?

In particular:

- Bbox question:
For statistical requests, is it necessary to include the BBox in the request? If I also send this data (BBox) is the PU used less?

- ResX ResY question:
To have precise statistical data relating to very small areas, like the ones I use, max 2500 m2 (square meters), how is it best to set the resX and ResY values?
I leave the default but I’m not sure it’s the best solution because I don’t understand how this data can influence the statistics (I don’t have to get the image).
So, which parameters of resX and ResY do you recommend?

- PU question:
Is there a way to know how much PU a single request consumed? So I can test and figure out how to consume less PU.

Thank you very much to anyone who wants to help me or provide me with some precise links where I can find information.
Greetings everyone.
HI,

-------------------------
-------------------------

The requests look like this:

import requests

url = "https://services.sentinel-hub.com/api/v1/statistics"
headers = {
"Authorization": "<by auth>",
"Accept": "application/json",
"Content-Type": "application/json"
}
data = {
"input": {
"bounds": {
"geometry": {
"type": "Polygon",
"coordinates": d

13.4990155251312,45.980966589012375],813.499047427833514,45.98096306634004],013.499076207665873,45.980952843147996],413.499099047450867,45.98093692015681],913.499113711471212,45.980916856023825],
13.49911876431732,45.980894614769475],113.499113711392582,45.98087237352394],313.499099047323647,45.98085230941404],313.499076207538648,45.98083638645138],313.499047427754881,45.98082616328244],
13.4990155251312,45.980822640618925],413.49898362250752,45.98082616328244],113.498954842723753,45.98083638645138],313.498932002938755,45.98085230941404],313.49891733886982,45.98087237352394],
13.498912285945082,45.980894614769475],113.49891733879119,45.980916856023825],513.498932002811534,45.98093692015681],913.498954842596529,45.980952843147996],413.498983622428888,45.98096306634004],
13.4990155251312,45.980966589012375]
]
]
}
},
"data":
{
"dataFilter": {},
"type": "sentinel-2-l2a"
}
]
},
"aggregation": {
"timeRange": {
"from": "2017-07-01T00:00:00Z",
"to": "2023-05-04T23:59:59Z"
},
"aggregationInterval": {
"of": "P1D"
},
"width": 512,
"height": 513.666,
"evalscript": "//VERSION=3\nfunction setup() {\n return {\n input: {\n bands: \n \"B04\",\n \"B08\",\n \n \"dataMask\"\n ]\n }],\n output: \n {\n id: \"myData\",\n bands: \"NDVI\",\"B08_NIR\",\"B04_RED\"]\n },\n \n {\n id: \"dataMask\",\n bands: 1\n }]\n };\n}\n\nfunction evaluatePixel(samples) {\n let ndvi = (samples.B08 - samples.B04) / (samples.B08+samples.B04);\n return {\n myData: ndvi,samples.B08,samples.B04],\n dataMask: dsamples.dataMask],\n \n };\n}\n"
},
"calculations": {
"default": {}
}
}

response = requests.post(url, headers=headers, json=data)

Hi,

I’ll be happy to help with answering your questions regarding the Statistical API.

  1. No, you are not required to include the BBox, so you can just use your input geometry instead.
  2. Regarding the X and Y resolution that you are using. To optimise your request I would recommend using 10m resolution as you are using the Red and NIR bands from the Sentinel-2 L2A data collection and this is their native resolution. Importantly, if it is not set at this then any resampling (upsampling or downsampling) would introduce bias into your results so when in doubt keep to using the native resolution!
  3. Lastly, regarding the PUs that your request uses, there is not currently a service that can do this automatically, however, I can direct you to our documentation that should help you calculate the PU costs approximately.

Lastly, because you are requesting data from long time periods (2017 to the present day), it is probably more appropiate to use the Batch Statistical API.

The Batch statistical API is only available for enterprise users. If you don’t have an enterprise account, and would like to try it out, contact us for a custom offer.

Hope that this information helps you out, and if you have any other questions then let us know 👍


Hi,
I really can’t find the right words to thank you!!! You were very kind and super competent!

The information and advice you gave me are very important to me, as it is the first time that I am developing programs that receive satellite data.

I hope one day to be able to return the courtesy and kindness you have shown towards me.

Thanks again so much.

Ciao


Reply