Hi!

With this script, large patches of complete black usually indicate water or no-data. It’s possible there was no Sentinel-2 data available on the day you checked. Please share the date and the area of interest you looked at, so we can make sure what happened. If you’re checking it in EO Browser, you can simply paste in the URL.

This is an example of fires in southern Turkey on July 30, where the script looks fine.

• EO Browser link

Hi! thanks for your reply. I tried for yesterday. However, the link you provided is very helpful to me, thank you very much. Can you explain to me what the color “purple” means in the URL that you sent? I understand that the color red shows fire, but did not understand what means “purple”. Thanks in advance.

here my link:
EO Browser URL

Yes, on August 1, there’s no Sentinel-2 image available for this area.

The purple seems to be an overestimation. Some agricultural fields are highly reflective in this area, making them look white on a true color image. White means they are highly reflective in all three, red, green and blue light. If you check only the Inverse calculation that the author is using for cloud classification, these fields have very bright values, even more than the smoke. The purple color is a result of the RGB bands in the output, which are all reflected by the fields. Checking the color (it’s 6768FF), the values of the RGB channels are (R:103, G:104, B:255). Red and green give approx. the same amount of color, but blue gives much more due to being multiplied by 20. So the purple is due to B04, B03 and B02 mixing. Here is a bit simplified script, and if I increase the B04 values, you see the effect in these overestimated areas becoming brighter pink.

In conclusion, the script overestimates cloudy areas in these fields, resulting in purple due to mixing of RGB bands in the output, which are all reflected by the fields. You can ignore them, and focus on purely blue (smoke) and red (heat).