Variance map outside fiber traces is very suspicious

Upper panel shows cutout from the science image from calExp-FA000031r1.fits and lower panel shows cutout from the variance image. As we can see, variance image is very different within fiber traces and outside fibertrace. I doubt that this is wanted behavior? It is creating problems for Keigo Nakamura when he wants to analyze PSF images in 2d, as the variance image outside fiber traces is very unrealistics.