Wednesday, July 9, 2008

Don't Clip Your Data!

I'm at the GBT, and just had a nice long chat with Crystal Brogan (who was very helpful when she reeally didn't need to be). I'm gonna list a few tips she suggests.

Here's one: DON'T CLIP!! She said that if you clip your data based on amplitude (or phase), what you're really doing is masking the lower-level bad data. A baseline is probably all bad if it has quite a few high points, and you don't want to just flag the really high stuff, you want to flag it all. If you clip the high stuff, it will be very hard to ever identify that whole baseline as bad. You'll essentially be 'losing' bad data. I know CLIPing is sometimes tempting, especially at the GMRT, but I can see that it's an especially bad idea in this case.

1 comment:

rob said...

Clipping should be safer on AMP DIFF or AMP V DIFF...