The numbers you quote are also not what's written there. He says "In Sevastopol, the vast majority voted for union, with a turnout of 50-80%. In Crimea [i.e. the rest of peninsula - not Sevastopol], according to varying sources, about 50-60% voted for the union, with a turnout of 30-50%".
It's not clear to me whether "Crimea" does or doesn't include Sevastopol in this text, but if at most 60% voted "yes" out of the at most 50% of the population that went to cast their vote, that's at most 30% of eligible voters who voted for the separation in whatever region they're talking about, compared to the 80% of eligible voters from the official numbers. You don't consider this a serious discrepancy?