Sentinel 5P Aerosol Index outside range

[links]

Hi, I am trying to download Sentinel5P data but I noticed the values for layers AER_AI_354_388 and AER_AI_340_380 are outside the normal range specified here. It is reported that the minimum value for these layers is -1 but I often encounter values lower than this threshold. I am downloading the data using Python SDK in this way

request = SentinelHubRequest(
                evalscript=evalscript,
                input_data=[
                    SentinelHubRequest.input_data(
                        data_collection=DataCollection.SENTINEL5P,
                        time_interval=(ts, ts),
                        mosaicking_order=MosaickingOrder.MOST_RECENT,
                        upsampling="BICUBIC"
                    )
                ],
                responses=[SentinelHubRequest.output_response("default", MimeType.TIFF)],
                bbox=city_bbox,
                size=city_size,
                config=config,
            )
            img = request.get_data()[0]

where evalscript is the following

function setup() {
    return {
        input: [{
            bands: ["AER_AI_354_388", "dataMask"],
        }],
        output: {
            bands: 2,
            sampleType: "FLOAT32"
        }
    };
}

function evaluatePixel(sample) {
    return [
        sample.AER_AI_354_388,
        sample.dataMask
    ]
}

the same for AER_AI_340_380 . I very rarely noticed some negative values also for HCHO layer which uses the exact same script.
Am I doing something wrong ?
Thanks in advance for the help

[chung.horng (Chung)]

Hi @links ,

Your script looks fine to me. As mentioned in the documentation, the Typical Range indicates what values are common for a given band and unit, however outliers can be expected.

I randomly downloaded an original file from Copernicus and the histogram did indicate that there’re values lower than -1 (Fig 1).

Fig 1

[links]

Hi @chung.horng, thanks for your reply.
I understand that Typical Range indicates common values but negative values do not make so much sense to me since they should be concentration values in mol/m^2. I would then consider those values as wrong acquisition and discard them, what do you think?

In addition, I noticed that the tile downloaded with my script and the values reported in EO Browser are different. More in detail, considering the metropolitan area of Milan on 2018-04-30 for some regions my downloaded tile has values lower than -1 while EO Browser shows values around 0.5 in the same area.
For this reason, I am still forced to think there is something wrong with downloaded data, or the alternative could be that the EO browser does some post-processing before visualisation. What do you think about this?

Thanks again for your support

[chung.horng (Chung)]

Hi @links ,

Could you provide your aoi and the timestamp which shows the discrepancy?

[links]

Hi @chung.horng,
my AOI is ( 9.0408867, 45.3867381, 9.2781103, 45.5358482 ).

Regarding AER_AI_340_380 and AER_AI_354_388 layers the timestamp in which I observed the discrepancy between my tile and EO Browser is 2018-04-30T10-27-52.

For what concerns HCHO layers I have to check the timestamp in which I obtained negative values but I can easily discard them as negative concentrations are for sure due to a noisy acquisition.

Thanks for your support