Coordinate alignment mismatch

[amiller448]

I have created a merged h5 using data from a single site that was deployed twice with a <10 minutes gap between each deployment and attempting to remote reference with another site. I am running into the logger error Coordinate alignment mismatch with reference to this previous issue. This continues for all decimation levels and creates unrealistic TFs.

As I have merged two datasets is there something more that needs to done to use a remote reference for creating the transfer functions?

[kkappler]

Hi @amiller448

From issue #228 it looks like this has something to do with time axes not matching exactly between local and remote stations. Can you paste the output from the logger? In the code I see the following lines:

logger.error("Coordinate alignment mismatch -- see aurora issue #228 ")
matches = X.time.values == RR.time.values
logger.error(f"{matches.sum()}/{len(matches)} timestamps match exactly")
deltas = X.time.values - RR.time.values
logger.error(f"Maximum offset is {deltas.__abs__().max()}ns")

so it would be good to know what those logging messages say.

Also, is there a way that you can share the mth5 file, the processing config and the script/commands/notebook that reproduce the error?

I have started on a patch for this (if it is what I think is causing it) and created a PR to address this issue. You can try pulling the fix_issue_383 branch, but there is a caveat -- that branch is pointed at the features branch which has not yet been released, so depending on the urgency, perhaps waiting for the (imminent) release of features branch will be better.

The patch currently does not fix the problem per-se, but will identify it better and raise a ValueError during the merging of the local and remote data if the time axes differ. A future patch could try forgiving a slight offset and forcing the time axes equal ... if that is something we want to consider doing.