Hi,
I’m a beginner in satellite image analysis and interested in snow cover data.
I want to reproduce the NDSI images displayed in EOB in ArcGIS PRO, as the DL images don’t contain values. But unable to do that. I couldn’t reproduce it using just the usual NDSI formula, so I looked at Javascript and it seems that on the EOB it uses values such as NDGR and bRatio to identify clouds. However, even when this was executed, many areas were identified as clouds.
I would like to know how to reproduce this. Is it possible to reproduce this?
Firstly, the NDSI script you see in EO Browser is just a visualisation script, therefore, the values you see in the output do not actually mean anything for scientific analysis. Therefore, you need to actually simplify the script a little for quantitative analysis. You can find an example of this here.
And to answer your question about cloud detection, unfortunately this is an inexact science and you will likely always have to live with a trade off between under or over detection of cloudy pixels.
Note that the output includes the sampleType parameter: output: { bands: 1, sampleType: "FLOAT32" }. I encourage you to read about this more in this blog post by my colleague @maxim.lamare
If you need any more help, then please reach out in the thread
Thank you for your reply. Sorry for the late reply.
OK, understand that there is no scientific basis for the NDSI images that can be downloaded from EO Browser. I will perform the analysis using the values I calculated myself.
