Strange phenomenon in remote sensing index NDWI

Question

I'm mapping water bodies in my region with Sentinel 2A Level 2-A imagery. I used ArcGIS Pro Raster Calculator to take (Band 3 - Band 8) / (Band 3 + Band 8) to get Normalized Difference Water Index. The process ran smoothly and at 0.05 and greater values, I can see all rivers, ponds, and storm basins.

However, there is a man-made basin that is used for recreation (jet skiing, weddings, concerts, etc.) that has me confused. The pixel values for only this basin range from -0.95 to -0.98.

What could cause that? Concrete-lined bottom? Total overgrowth of algae? Drained? I created a true color composite and the basin looks very dark, similar to nearby canals and ponds.

Answer 1

Thanks for the question. I think it's interesting, but not suited to GIS Stack Exchange. It might be better answered in the Physics stack exchange.

Really what you're asking is how the interaction of light with the water column of a particular water body can change the Green - NIR ratio enough to make it be classified separately from other water bodies.

A high negative number shows a large difference between bands, and a dominance of the NIR channel. We know vegetation reflects NIR well. My guess is your issue has to do with vegetation.