To my knowledge there are no tools that attempt county-level resolution of Y SNP ancestral location. SNPs were not known in the era when national census-tasking provided location data for surnames, and even today the data for direct association of Y DNA with ancestral county is very sparse (for example https://www.familytreedna.com/groups/british-isles/about ).
However, men who do Y SNP testing at Family Tree DNA (= FTDNA) may optionally enter their surnames, which FTDNA makes accessible in their comprehensive tree of Y SNPs. From historic census data we can associate surnames with counties. All that remains is to make the connection: Y SNP associated surnames associated counties, which this tool does with appropriate weighting and averaging.
We would expect that the SNP map for "Y Adam" (try A-V221) should be evenly colored since this is every man's common paternal ancestor. Yet it does not come out that way because of the random nature of surname tagging: the high variance of small numbers is apparent for small counties. Therefore all SNP maps are flattened by dividing all values by the median county scores seen for A-V221. While such fudge factors should usually be avoided, in this case the flattening makes weak but interesting patterns visible. Try R-L21 with R-U106 for example: the difference between Bell Beaker R-M269 and Anglo-Saxon R-M269 is subtle but consistent.
Our expectation is that SNP localization may work at a regional level (e.g. Scotland vs Wales vs Cornwall/Devon) but may fail for most surnames since Leslie et al (Nature 2015 https://www.nature.com/articles/nature14230.pdf ) showed that populous central England is well mixed. This tool provides two ways to gauge the validity of any SNP mapping: first, check the number of surnames with map data under the SNP (i.e. surnames in blue bars in the left column under Details). If there is only one such surname (for example Lynch for R-BY4102), then we know that the map is based solely on that surname and the pattern cannot be older than when surnames arose in the 12th-13th centuries. But if there are several such surnames (for example Duncan, Doherty, McKee, Boyle for R-S673), then we can infer a common ancestry for these branches of these names and have more confidence in their co-location.
The second gauge of SNP location validity is the percent geographic similarity to the average of all surnames. As we move up the SNP tree (back in time), at some point we sweep in everyone's ancestry and there cannot be any geographic differentiation. For that purpose this tool uses SNP A-V221 which was formed 133,000 years ago, so that anywhere outside of Africa the geographic distribution for A-V221 is guaranteed to be all-inclusive (and so uninformative). Thus any SNP with similarity approaching 100% of A-V221 is too inclusive and too well mixed to have an interesting map pattern.
SNP patterns are especially interesting if the SNP was formed well after one of the major bottlenecks, specifically after the starburst ~4000 years ago under R-M269. This topic will eventually get more discussion and illustration in one of my Research Reports.