Here is one of the most complex questions if you want to be precise. In fact, it is nearly impossible (and in the end, not that useful) to have a razor-thin margin on that:
– The real number of residents changes each day in a large city, as there is always someone coming or going. The statistic use of knowing that this weekend there will be 138 (not 137, not 139) college students visiting their moms to go back to college on Monday is rather relative…
– The official sources (census, inscription at City Hall) are published for a given day, in the best case each year. So they are accurate for… a day.
– The real number of residents in a city can depend on the ratio of cheaters. For instance, when residing in a city can allow you the access to a parking slot for residents, or a school for your offspring, it is not unusual to see families move to a peripheral municipality, while trying to stay registered as residents in the central city. Maybe a little fraud, but hard to detect and correct (central municipalities, as all, receive money in proportion to registered residents, and on the other side sprawling municipalities derive most income from building and consumption).
– Even if you are sure that your citizens never cheat, if they live in a metro area their realities go beyond municipal border.
Besides, beyond a certain moment, knowing the precise figure has a marginal use; the use citizens make of resources (water, electricity, transportation…) do not depend just on the number of residents, but rather on their consumption patterns, which are more complex to know, and even more to predict, and here the systematic data about recent resource use can be of help.
It is important to have a trustworthy reference figure for the city population, but even more important is to know the structure of these populations. For instance, calculating the need for schools depends more on the number of compulsory school age kids (variable, but not that much during a given year) than on the overall population, and the same goes for other issues. The problem of data falsified by residents to profit reappears, and so, again the relevance of the final figure (kids in schools) rather than the official population figure.