The calibration of the MWRP drifted starting on September 21. The drift caused a negative
LWP. On 10/21 I modified the calibration of the 30 GHz channel in an attempt to correct
for the negative bias in LWP. The change in the calibration however caused an excessive
increase in the PWV retrievals. The original calibration values were reinstated on 10/29. I
then recomputed the temperature coefficients as well as the noise diode injection
temperature from tip curves and uploaded the new values on Nov 15 2010.
Reprocess the data from 9/21 to 11/15 with the newly computed coefficients.
Reprocessing completed 20110111.