I've cleaned up the production database for Great Lakes Monitoring if you guys would like to take a look. I haven't deployed this to production, but I've uploaded it to both the seagrant production server at /home/angelo/2016-02-01-cleanup.sql.gz

I had a hard time figuring out what was different between this database and the dev database, but managed to make some progress. Here is what I did.

First, Gene and I removed all of the sensors, streams and datapoints created on dev after October 2015. There were a lot in there and Gene wanted to clean up the parameters.
The dev database had these counts before:

"sensors": 150,
"streams": 5498,
"datapoints": 716243

After we cleaned them up, it had these counts:

"sensors": 131,
"streams": 4034,
"datapoints": 689539

Then I pulled down a copy of the production database and removed streams without datapoints and sensors without datapoints.
When I compare the cleaned up production database to the cleaned up dev database, I still see a discrepancy and I'm not sure the cause of it. Here is production:

"sensors": 144,
"streams": 4972,
"datapoints": 958518

And here is the cleaned up dev database again:

"sensors": 131,
"streams": 4034,
"datapoints": 689539

The sensors are accounted for: Marcus added NOAA and WQP sensors, exactly 13 of them.
However, he only created one stream for each sensor, so I'm not sure why we have more than 900 extra streams on the production database.
I then counted the number of datapoints that Marcus created for NOAA and WQP and got 122,755 datapoints, but the production database has 268,979 more datapoints than the dev server, so there are 146,224 extra datapoints on production.

That may be fine, it is possible that we added datapoints to existing sensors, but I don't have an easy way of knowing for sure.