Trimming wide confidence interval segments of a loess curve

Question

[lefthandgergo]

In one of my analyses, I have loess curves with confidence intervals. I would like to keep only those segments of the curve, where the confidence interval is narrower (between the vertical black lines). Is there any accepted way to determine an optimal trimpoint for loess curves, apart from checking them by plain eye? I searched for it, but couldn't find any technique in the literature. Thanks!