Dear Mark, Thank you for your DQE evaluation of CTD data collected along WOCE sections P14S and P15S. We considered each of your suggestions and the following is an itemized explanation of what we did or didn't change in our data files, as well as answers to your questions. Kristy McTaggart and Greg Johnson *************************************************************************** STATION SUMMARY FILE (.sum) Stations 21 and 77 should be listed as cast 1. The .sum and .ctd files should be corrected. We've corrected our files here. The uncorrected sounder depth at the bottom of the cast for stations 44 and 55 may appear erroneous. However, these are not typos. They are the values calculated from the ship's PDR during acquisition. The bottom at station 44 in particular was noted to be strongly sloping. We did not change these values in our files. The PDR sound speed used for sounder readings was 1500 m/s. The readings were not corrected for transducer depth below the waterline. The depth of the transducer would've been about 5.5 +/- 0.6 m. We would prefer to use the PDR depths as listed and correct them using Carter's tables so that they serve as independent measurements and can be used as a check on CTD pressure. SALINITY 'Scatter of salinity residuals' There is an incompatibility between the General Oceanics rosette sampler and the Sea-Bird 911plus CTD system that generates a spike in the data stream at the moment a bottle is confirmed as tripped. Because of this, upcast CTD burst data had to be averaged prior to the bottle confirm bit. Two-second averages were chosen over a longer interval because the CTD operators did not always let the package sit at bottle depth for at least 10 seconds before firing the rosette. Hence no changes were made. 'Biasing of CTD salinity data for individual stations' Of course one can seemingly make a (very slight) improvement in the CTD-bottle residual statistics by allowing more degrees of freedom in the fit as the DQE has suggested (that is, breaking up the fit into small station groupings). One could get the best statistics by individually fitting each station to its bottles, but most experts would argue that this would be a bad choice, because one would not be taking advantage of the CTD calibration as a way to average out station-to-station bottle salinity noise. We believe that the SBE-9/11 CTD conductivity slope drifts gradually, and is actually more stable than the day-to-day fluctuations in the autosal- inometer salinities owing to small temperature drifts in the laboratory and the fact that severe budgetary constraints on these cruises forced us to economize even on such things as standard sea water. We suspect that the "biasing of the CTD salinity data" mentioned in the DQE evaluations is actually noise in the bottle data. Somewhat suspicious is that the station groupings recommended by the DQE of the correct size (most often 3-5 stations per group) that they could easily be owing to daily drift problems in the autosalinometer. For our original calibrations we deliberately chose to model the conductivity slope adjustments of the entire data sets for P14S/P15S and P18 using 4th-order polynomial functions of station number to average out bottle salinity noise. We did this because we saw no obvious jumps in the CTD calibration for either cruise, just gradual drifts. Statistical support for our philosophy over that of the DQE is given by the following exercise: The 2oC potential isotherm is well within the oldest Pacific Deep Water, and has some of the tightest Theta-S relation- ships in the Pacific Ocean (and probably the world). For both P18 and P14S/P15S, we looked at the absolute values of station-to-station changes in CTD salinity on Theta=2.0oC (Figure 1) for our original calibration, creating a histogram of station-to-station differences for each cruise in 0.001 bins. We then applied the DQE's suggested ad-hoc calibrations for smaller station groupings to the data and conducted the same analysis. When the histograms are differenced (Figure 2), one can see that the Theta-S relations at 2oC after the DQE's corrections are noisier for both cruises. For P18, after the DQE's suggested correction there are four less station pairs in the 0.000 difference bin and one less in the 0.001 difference bin whereas there are three more in the 0.002 difference bin and two more in the 0.003 difference bin. For P15S/P15S there are four less stations in the 0.000 difference bin after the DQE's suggested correction, with one more in the 0.001 difference bin and three more in the 0.002 difference bin. Since the DQE's "corrections" actually introduce more noise in the CTD Theta-S relation at 2oC than our original calibration, we decline application of them. The small groups do not improve the calibraiton, they degrade, perhaps by introducing autosalinometer drift noise. Regarding suspicious CTD salinity data listed in Table 3, no changes were made to any profile data (see above) nor flags associated with "transient/ despiking errors". As for CTDSAL values in the .sea file for station 127, we agree that they should be flagged as 3 for samples 202 to 214. Also, BOTSAL flags for samples 209, 210, 213, and 214 should then be changed to 2. 'Problem salinity bottle data' Excluding stations 19, 49, 117, and 164 bottle salinity values from the calibration of this data set as a whole would not significantly change the fit as we have done it, thus we didn't make this adjustment. OXYGEN Quality flags should be ammended as suggested in Table 4. However, stations 8, 10, and 135 will not be recalibrated individually as they are among the first casts with a new sensor module. As a rule, the first few casts with a new module are problematic, and this cruise was no exception. TEMPERATURE The very spikey temperature structure between 100 and 300 dbar at station 43 is also seen in salinity and has been identified as Antarctic Intermediate Water interleaving at the front. It is also seen at adjacent stations 42 and 44. Nothing should be done to this profile. Temperature spikes listed were examined but not changed. Neither were their flags changed. DESPIKING AND INTERPOLATION Interpolated temperature and salinity data are the result of processing programs and not instrument or electronic problems. In program DESPIKE salinity profiles are viewed and interactively despiked using linear interpolation. Conductivity, theta, and sigma-theta are recomputed for the interpolated records. Only the salinity quality flag is ammended to 6. In program DELOOP Brunt-Vaisala Frequency squared (N^2) is computed at the mid depths and bracketed between two vectors, one padded with zeros at the surface and one padded with zeros at depth. If the first and second points of a -N^2 fail the criteria (<=-1e-05), then temperature and conductivity are linearly interpolated and salinity, theta, and sigma- theta are recomputed. The quantity of interpolated points is large because we were working with a large package off the stern of the ship, often in the Southern Ocean. Hence, there was a lot of wake problems. As for the filled surface records flagged as 7, we maintain that this is more useful than leaving flagged bad or questionable data or removing the data entirely. It should be noted in the documentation that all data in the top 15 dbar with a flag of 7 should be regarded as questionable. DENSITY INVERSIONS Density inversions listed in Table 5 were examined and salinity quality flags were changed to '3' for the following records. Station Pressure Station Pressure 8 5- 7 108 4 10 1- 7 109 1 20 1- 3 110 1 22 1- 5 111 1 45 1- 8 112 1 49 1- 7 113 1- 3 54 7 117 1- 6 57 1 120 1 60 5- 6 121 1 64 7- 8 124 1- 3 68 1 125 1- 3 69 1-14 126 1-13 70 3, 5 131 3,5,6,10 71 6 132 1- 9 78 1- 9 134 1-3,6 82 1- 4 135 1 83 7 136 1 84 1- 2 139 8 85 4 140 4, 5 86 1 143 1- 3 87 1 144 1 88 3, 4 146 1- 6 89 3, 4 148 1- 3 90 4, 8 152 1- 2 91 1- 4 153 1- 2 99 1- 2 154 1- 3 101 1,3,7 155 1-15 102 6 157 1- 4 105 1- 3 159 1- 6 106 1-3,6,7 160 1-12 107 1-2,8,11-13 162 1-13 165 1- 3 167 1- 3 169 1- 7 170 1- 3 174 1- 3 176 1- 4 177 1- 3 180 1- 3 181 1 182 1- 2 DOCUMENTATION Again, the PDR sound speed was 1500 m/s, and the readings have not been corrected for transducer depth (5.5 +/- 0.6 m) below the waterline. The criteria used for despiking is explained above under DESPIKING AND INTERPOLATION.