Use this doc for a description of vCenter adapter metrics:
Look to the bottom of the pdf for metric details.
Usually with the "usage capacity" metrics for disk and network, I've observed it pulling from the high watermark of the usage levels. Obviously there is an elaborate formula for it, but that's the general idea. Honestly, the usage capacity is leveraged for planning/cap planning and not used during your 'average' problem drilldown. I personally wouldn't pay it much mind during a drilldown/metric view unless for some reason you thought your current usage was approach the "capacity" and might hit the usage ceiling.