Is it me, or does there seem to be more variation in the 2000-10 data than the 2011-16 data for the outer areas of SE Michigan and SE Wisconsin? Does Census revise the population estimates after the census to reflect the next Census or something? Or is Census just estimating things by county in the smaller jurisdictions until the 2020 census?
The annual population estimates for states and counties are based on births, deaths, and migration figures. The census bureau doesn't have these data for cities and towns, but they do make an annual estimate of housing units for municipalities based on building permit data. The annual population estimates for municipalities take the county estimate and then distribute it among the municipalities based on housing units. But if I understand the method correctly, they don't take into account changing rates of occupants per housing unit - so if one town has a new subdivision marketed towards young families with around 4 people per new house and another town has one marketed towards seniors with 1-2 people per house, this won't show up in the annual estimates.
After the census, the annual estimates for the cities and towns are revised for the previous decade based on the actual census data.