invalid UTF-8 byte sequence detected

Based on a couple of other posts, I think I'm out of luck, but I'm hoping
something might have changed recently.

I'm loading a very high volume of data with COPY using libpq - about 100+
million rows per day. The problem is that the COPY sometimes aborts with

invalid UTF-8 byte sequence detected

I'd prefer not to add to the overhead by pre-validating every string, since
PostgreSQL validates it already. Is there a way to get the server to
replace invalid characters with something like blank instead of generating a
fatal error?

Wes