TheIn his Classification of Place Names, George Stewart includes "false description", a subgroup of his category 1, "descriptive names":
Actual false description is rare. Most of its examples would be better classified under euphemistic names. Others are to be explained as incident names, that is, the original namers observed the place under unusual circumstances and their name perpetuates these circumstances, and does not describe the ordinary nature of the place
While this seems to be a valid answer to the question, it's so straightforward it's boring...
More interestingly, the terms "transparency" and "opacity" can be applied to toponyms, as Radding and Western did in their article "What's in a name? Linguistics, geography and toponyms". They use the example of Newcastle in England:
... few people today associate the city in northeastern England whose name is "Newcastle" with any castle, new or otherwise [...], therefore, the toponym is now close to opaque.
However, Helmar raises the valid point that the "opacity" of an opaque descriptive toponym can be due to the fact that the descriptive properties of the toponym aren't used or useful anymore (like in the Newcastle example), and not only when the description is wrong. So the term "opaque descriptive toponym" isn't a perfect answer to the question, but I'd like to keep it in this answer nonetheless. The fact that transparency/opacity is a continuum and not a dichotomy like "true/false description" reflects how place names can be connotative (indicating attributes of the named object), denotative (merely identifying the named object), or something in between (see the interesting discussion in Nicolaisen, "Are there Connotative Names?")