FInding "corrupt" values in UTF-8 tables (regexp question, I think)

I'm noticing that some of my data has been imported as junk text:

For instance:

klciã«"

What would be the SQL to find data of this nature? My column can only
have alphanumeric data, and the only symbols allowed are "-" and "_",
so I tried this regexp query:

select id, t_code
from traders
where t_code ~ '[^A-Za-z1-9\-]'
limit 100;

But this starts to return values such as "181xn-807199" which is valid
as per the above regexp? Also, when I try to include the underscore,
as follows...

select id, t_code
from traders
where t_code ~ '[^A-Za-z1-9\-\_]'
limit 100;

This gives me an error: "ERROR: invalid regular expression: invalid
character range".

What am I missing? Does this have something to do with erroneous
encodings? I want my data to be utf-8 but I do want to find it with
latin1 queries when the text in columns is supposed to be only latin1
characters! Or is "a-z" in utf-8 considered different from "a-z" in
latin1?