change in the NDVI value when compared to sentinel hub

[krishnaglodha]

Hi, I'm trying to create NDVI based on sentinel-2-l2a , bit I see difference in values on pixels when compared with sentinel hub

Here is the image from sentinel hub

we can see that it ranges from 18 to 61 i.e. 0.07 to 0.23 ( divided by 255 )

and here is the image from planetary

where values are ranging between 0.04 to 0.15

my code looks like

catalog = pystac_client.Client.open(
    "https://planetarycomputer.microsoft.com/api/stac/v1",
    modifier=planetary_computer.sign_inplace,
)
time_range = f"{start_date}/{end_date}"
collection = "sentinel-2-l2a"
geometry = polygon["features"][0]["geometry"]
search = catalog.search(
    collections=[collection], intersects=geometry, datetime=time_range
)  # bbox=tas_bbox   
print(search.get_all_items())
data = load(
            search.items(),
            geopolygon=geometry,
            chunks={},
            resolution=1,
            projection="epsg:3857",
        ).isel(time=0)
dc = data.rio.clip([geometry], crs="epsg:4326")
index_xarray = ms.ndvi(dc.B08, dc.B04)
index_xarray.rio.to_raster("planet_scope.tif")

[TomAugspurger]

.Looks like a few variables in your snippet aren't defined, so I can't reproduce it (start_date, end_date, ms)

Do you know offhand whether the values from Sentinel Hub are L1C or L2A, and which processing baseline it's using? It might be better to compare just the raw values (from before the NDVI calculation).

[krishnaglodha]

here are the variables

polygon = {
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "properties": {},
      "geometry": {
        "coordinates": [
          [
            [
              79.06776796168413,
              19.043044385618458
            ],
            [
              79.06776796168413,
              19.041761558038218
            ],
            [
              79.06973122024931,
              19.041761558038218
            ],
            [
              79.06973122024931,
              19.043044385618458
            ],
            [
              79.06776796168413,
              19.043044385618458
            ]
          ]
        ],
        "type": "Polygon"
      }
    }
  ]
}

start_date = '2024-01-01'
end_date = '2024-01-06'

[krishnaglodha]

PS: I also tested out result in comparison with agromonitoring.com , and got similar differences

vs microsoft

[krishnaglodha]

.Looks like a few variables in your snippet aren't defined, so I can't reproduce it (start_date, end_date, ms)

Do you know offhand whether the values from Sentinel Hub are L1C or L2A, and which processing baseline it's using? It might be better to compare just the raw values (from before the NDVI calculation).

when we download raw data from pc the range is between thousands, whereas sentinel-hub data range is betweek 0-255, what might be going wrong ?

In terms of data, we are using L2A everywhere

[TomAugspurger]

when we download raw data from pc the range is between thousands, whereas sentinel-hub data range is betweek 0-255, what might be going wrong ?

That might be to the baseline or units I mentioned in #350 (comment).

[krishnaglodha]

when we download raw data from pc the range is between thousands, whereas sentinel-hub data range is betweek 0-255, what might be going wrong ?

That might be to the baseline or units I mentioned in #350 (comment).

how do we use this units ?

[krishnaglodha]

I tried replicating same results with 'https://earth-search.aws.element84.com/v1' and surprisingly I got correct result ( similar to sentinel hub ) with earth search STAC link.

[TomAugspurger]

how do we use this units ?

Sorry, I don't understand the question.

I'm not an expert, but my understanding is that bands like "B04" are provided as "digital numbers" and are typically in the range 0-4000.

[krishnaglodha]

I tried replicating same results with 'https://earth-search.aws.element84.com/v1' and surprisingly I got correct result ( similar to sentinel hub ) with earth search STAC link.

@TomAugspurger , what is your comment on this ?

[TomAugspurger]

I’m not sure, I haven’t used that data before.

…

________________________________ From: Krishna Lodha ***@***.***> Sent: Monday, May 27, 2024 10:19:13 AM To: microsoft/PlanetaryComputer ***@***.***> Cc: Mention ***@***.***>; Comment ***@***.***>; Subscribed ***@***.***> Subject: Re: [microsoft/PlanetaryComputer] change in the NDVI value when compared to sentinel hub (Issue #349) I tried replicating same results with 'https://earth-search.aws.element84.com/v1' and surprisingly I got correct result ( similar to sentinel hub ) with earth search STAC link. @TomAugspurger<https://github.com/TomAugspurger> , what is your comment on this ? — Reply to this email directly, view it on GitHub<#349 (comment)> or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAKAOIXUF4WHAHVJFVIFCM3ZENFHFBFKMF2HI4TJMJ2XIZLTSOBKK5TBNR2WLJDUOJ2WLJDOMFWWLO3UNBZGKYLEL5YGC4TUNFRWS4DBNZ2F6YLDORUXM2LUPGBKK5TBNR2WLJDUOJ2WLJDOMFWWLLTXMF2GG2C7MFRXI2LWNF2HTAVFOZQWY5LFUVUXG43VMWSG4YLNMWVXI2DSMVQWIX3UPFYGLLDTOVRGUZLDORPXI6LQMWWES43TOVSUG33NNVSW45FGORXXA2LDOOJIFJDUPFYGLKTSMVYG643JORXXE6NFOZQWY5LFVEZTCNZUGAYTKMZUQKSHI6LQMWSWS43TOVS2K5TBNR2WLKRSGMYTENBYGUYTQNVHORZGSZ3HMVZKMY3SMVQXIZI>. You are receiving this email because you were mentioned. Triage notifications on the go with GitHub Mobile for iOS<https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android<https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>.

[brendancol]

@TomAugspurger @krishnaglodha For things that are specific to xrspatial.multispectral.ndvi happy to help as we have this issue that was setup..makepath/xarray-spatial#802

[jessjaco]

Just a thought, the other sources might be applying the offset automatically, see #134. Maybe try a date before the offset was added and see if the values match?

[mhungen]

I tried replicating same results with 'https://earth-search.aws.element84.com/v1' and surprisingly I got correct result ( similar to sentinel hub ) with earth search STAC link.

They seem to apply the offset to there COG assets. I ran into a similar problem and documented it. here

the confusing part is that they use the STAC raster extension to provide the values for offset and scale.

"raster:bands": [
    {
        "nodata": 0,
        "data_type": "uint16",
        "bits_per_sample": 15,
        "spatial_resolution": 10,
        "scale": 0.0001,
        "offset": -0.1
    }

However the offset is still "-0.1". IMHO it should be "0" since the values are already harmonized. but this is another story....

the deviations you have observed can also be caused by the other DEM used by PC

[ghidalgo3]

Closed due to inactivity, feel free to reopen if you would like to continue this discussion.