Critical Issues

• Much of the processing has come to depend upon the NetCDF CF conventions. The quality variables are uniquely specified from the conventions, as are standard names. The pre-deployment initializations require that standard names be present.

• The web updates depend upon the MATLAB m-file update_db_last_obs. This routine is only efficient because of a MySQL index created upon certain columns in the MySQL table processing_datatypes.

• The instrument_number attribute is only important for locating a file or variable within a deployment. When looking at the historical data, it is potentially misleading, as instrument ids for a variable at the same depth can change. For example, on the M0101 and M0102 deployments, we had SBE37 depths of 283 meters. That instrument had id 0. But for M0103, we had SBE37 depths of 1, 20, 50, 100, 150, 200, and 250 meters, having instrument ids of 0, 1, 2, 3, 4, 5, and 6 respectivitly. Obviously the 283 meter instrument on deployments M0101 and M0102 do not match up with the 1m instrument on M0103.