Table of Contents

INTRODUCTION.. 3

1. INSTRUMENTATION.. 5

1.1 Instruments and Materials. 5

1.1.1 Autosal instruments. 5

1.1.2 Plastic and glass bottles. 6

1.1.3 Substandard batches. 9

1.1.4 Data entry instruments. 9

1.2 Operators. 10

1.3 Maintenance. 11

1.3.1 Regular maintenance. 12

1.3.2 Repair 13

2. METHODOLOGY.. 14

2.1 Autosal Operation. 14

2.1.1 Duplicate measurements. 14

2.1.2 Bath and ambient temperature range. 16

2.1.3 Good Autosal operator habits. 18

2.2 Documentation. 19

2.2.1 Measurement Sheets / Run Log. 19

2.2.2 Written reports. 20

2.3 Data Processing. 21

2.3.1 Changes in processing programs. 21

2.3.2 Corrections. 22

3. STANDARDIZATION.. 24

3.1 IAPSO Standard Seawater 24

3.1.1 Standardization methods with IAPSO standard seawater 25

3.1.2 Autosal electronics drift 26

3.1.3 IAPSO batch comparisons. 30

3.2 Substandard Seawater 30

3.2.1 Substandard history. 32

3.2.2 Frequency of substandard measurement 36

4. QUALITY CONTROL. 38

5. CONCLUSION / SUGGESTIONS. 46

6. REFERENCES: Appendices. 48

 

 

INTRODUCTION

For over fourteen years, the Hawaii Ocean Time-series (HOT) project has been successfully building and maintaining a database for observing and interpreting physical and biogeochemical variability in oligotrophic waters around the Hawaiian Islands.  The HOT project has established an extensive compilation of data, collected at nearly monthly intervals, which are readily available to the public and to the scientific community (http://www.soest.hawaii.edu/HOT_WOCE/dataftp.html).  The Physical Oceanography (PO) component of the HOT project has been studying the relation of water mass variations to gyre fluctuations, determining the needs and methods for monitoring currents at Station ALOHA (A Long-term Oligotrophic Habitat Assessment) and developing a climatology of short-term physical variability (http://www.soest.hawaii.edu/HOT_WOCE/).  Thus, it is extremely important to maintain standardized methods for effective sampling and analysis in order to provide reliable data to the scientific community. 

 

Salinity sample measurements, one of the components for which the PO group is responsible, are extremely important in that 1) they provide primary calibration for CTD profiles and thermosalinograph data, 2) they comprise a database of water salinity characteristics at various depths in the water column for an extended period of time at Station ALOHA, and 3) they provide quality control on other Niskin bottle water samples for other measurements.  Standardized methods in sampling, analysis, documentation, and calibration are crucial in setting a context for extracting meaningful signals from a time-series that has been in effect for as long as the HOT Project.  In this report, we have reviewed the history of the HOT salinity database in order to quality-control sections of data with confidence parameters.  We have reviewed past and present methods and will recommend guidelines and standards for optimal procedures.  As a comprehensive history analysis, this report is organized into four parts: Instrumentation (Section 1), Methodology (Section 2), Standardization (Section 3), and Quality Control (Section 4).

 

The Instrumentation section (Section 1) presents the history of salinity sample measurement equipment and documentation of Autosal maintenance over the years.  We present changes in salinity measurement instruments and materials as well as changes in instrument operators.  We also stress the importance of documentation in the Autosal Run Log by the Autosal operator, including any physical, electrical, or chemical repairs done on the Autosal instrument, which should be entered into the Maintenance Database by the Electronics Technician. 

 

In the Methodology section (Section 2), we evaluate the overall methodology of the salinity bottle sampling and data analysis process.  Here we discuss which of the methods employed in the past have remained, been updated, or been discontinued, and ways in which we can improve organization and eliminate confusion in the future.  Some of the important procedures that need to be standardized include methods in Autosal operation, documentation, and data processing.

 

The Standardization section (Section 3) is a subset of the Methodology section in that they both relate to specific methods in the salinity measurement and analysis process.  The Standardization section, however, is specific towards standardization and calibration methods using IAPSO and substandard seawater (collected in a 50 liter carboy from 1000 m at Station ALOHA).  We also examine the significance of Autosal electronics drift throughout the years.  A close look at standardization methodology and Autosal drift may give clues to whether adjustments are needed for specific groups of data.

 

The Quality Control section (Section 4) includes the methods used to determine the quality of the salinity samples by flagging them as good, suspect or bad. We also include plots of salinity values for samples below 3000 dbar, to give an indication of the salinity variability in the deep and bottom water.

 

A final section with Conclusions and Suggestions (Section 5) contains a summary of this report, including recommendations for improving the salinity data quality.
1. INSTRUMENTATION    

1.1 Instruments and Materials

1.1.1 Autosal instruments

A timeline shows the salinity measurement instrument used for each HOT cruise (Figure 1).

 

 

Figure 1. Instruments used to measure salinity samples during the HOT project cruises 1 through 149. The horizontal axis indicates the cruise number.

During HOT-1 through HOT-26, salinity samples were measured by Ted Walsh at the Analytical Services Lab on the 5th floor of MSB using an AGE Minisal 2100 salinometer.  The “Salinity Measurements” section in the World Ocean Circulation Experiment (WOCE) Operations Manual (Stalcup 1991, Appendix A) recommends Guildline Autosal salinometers as being the only type of instrument that is capable of achieving the required WOCE accuracy for salinity measurements. 

 

The HOT project, originally a part of WOCE, purchased the Guildline Autosal 8400A salinometer (SN: 58296, UH Decal: 065QV) in 1990.  Prior to switching over to the Autosal 8400A, comparisons were conducted between Autosal 8400A and Minisal measurements using substandards and duplicate salinity samples from HOT-22 through HOT-26.  From HOT-27 through HOT-107, Autosal 8400A was used to measure primary salinity samples.  For the first six cruises (HOT-27 to HOT-32), the Minisal was used to measure duplicate salinity samples for further comparison (Kennan 1992, “Summary of Comparisons between Guildline Autosal and AGE Minisal,” Appendix B).  Around 1993, an older model of the Autosal 8400A (Autosal 8400, UH Decal: 916RF) was purchased by the Joint Institution of Marine and Atmospheric Research (JIMAR) and was given to the Roger Lukas group.  This Autosal 8400 was used to rummage for parts and as a troubleshooting guide when the Autosal 8400A encountered problems.  The Autosal 8400 was formally disposed of and removed from inventory in June 2001.  After HOT-80 (March 1997), an adjustment was made on the Autosal 8400A to reset the standardization knob, which reached its minimum setting value and could not turn any further.  After the standardization knob was reset, we continued to use the Autosal 8400A until HOT-107. 

 

The Autosal 8400B (SN: 63903, UH Decal: 924SP) was purchased in February 1999 using NSF shipboard equipment funding and has been used to measure primary and duplicate salinity samples since HOT-108 until present day.  On September 19, 2002, a comparison experiment between Autosal instruments 8400A and 8400B was conducted by Daniel S. Fitzgerald (see his experiment notes, Appendix C) after Autosal 8400A was repaired in the lab by the Electronics Technician.  He concluded that the Autosal 8400A is functioning correctly and that it is possible to utilize the Autosal 8400A in case of an emergency with the Autosal 8400B.

 

1.1.2 Plastic and glass bottles

Previous salinity reports and records suggest that while Ted Walsh measured salinity samples on the Minisal, salt samples were collected in 250 ml high-density polyethylene plastic Nalgene bottles.  Evaporation is the single most important factor that affects salinity samples in plastic bottles.  Though parafilm was used as additional sealer for plastic bottles, evaporation still occurred through the sides of the bottles.  Along with the switch to the Autosal 8400A, the group also started using 200 ml flint glass bottles.  The glass bottles, equipped with screw caps and Poly-Seal caps to prevent leakage and evaporation, meet WOCE specifications more readily than the plastic bottles.  The first record of glass bottles being used is on October 4, 1990 by Sean Kennan during his test of evaporation in plastic bottles using glass bottles as reference samples (Kennan 1991, “Evaporation of Plastic Bottles,” Appendix D).  Kennan reports that samples in plastic bottles not measured within 15 days after the cruise must be corrected by an evaporation rate of 1 mpsu/14 days or 0.071 mpsu/day.  Thus, all samples collected in plastic bottles should have been corrected if measured on or later than the fifteenth day after the cruise (due to our own threshold of making corrections for differences >1 mpsu). A review of these corrections is presented below.

 

The first cruise that utilized glass bottles for some casts seems to be HOT-23 (February 1991).  These samples were measured on both the Minisal and the Autosal 8400A and the results are summarized in a report by Sean Kennan (Kennan 1991, “Summary of Salinity Measurements from HOT-23 (2/1 to 2/6, 1991),” Appendix E).  After switching to glass bottles during HOT-24, the bottles were numbered in sequential order from #1 to #408.  When there were not enough glass bottles during a cruise, plastic bottles were used, mostly for thermosalinograph and duplicate samples.  HOT-82 shows new glass bottles (#501 and up) being used; however, plastic bottles were still used for thermosalinograph samples.  HOT-84 (June 1997) is the first cruise in which only glass bottles (#1 to about #700) were used for all casts and since then, only glass bottles have been used.  Since bottles should be replaced every eight to ten years according to the WOCE Operations Manual, the group started using a second set of bottles (#501 to #980) during HOT-140 (September 2002).   Table 1 below categorizes the bottle types used for sets of HOT cruises and whether evaporation corrections were applied to the respective sets of data.       

 

 

 

1.1.3 Substandard batches

Secondary lab-standard (substandard) seawater batches were used from the beginning of the HOT program in order to monitor drift of the Autosal within and between measurement sessions.  Substandard batches were first used sometime around HOT-23 (February 1991).  The seawater used for these batches was collected from 1000 m at Station ALOHA, stored in a glass carboy topped with mineral oil to prevent evaporation, and covered with a black plastic bag to prevent biological growth in the seawater.  Appendix G is a record of Sean Kennan’s calculations on how much mineral oil is needed for the glass carboy.  A maintenance record on July 16, 1991 (Valenciano and Rii 2003, Autosal History Maintenance Database Report, Appendix H) suggests that air pumps on the Autosal 8400A malfunctioned possibly due to mercuric chloride, which had been put into the substandard batch along with mineral oil to inhibit biological growth.  Use of mercuric chloride ceased starting batch #4 (December 1991).  Details on substandard seawater batch use are discussed in Section 3.2. 

 

1.1.4 Data entry instruments

After HOT-24 (March 1991), the group started using a ‘C’ program called ‘autosal’ to enter handwritten Autosal output data into the computer (Appendix I).  This interactive program, originally written by Janice Sato, automatically converted the Autosal’s electronic reading (2 x K15 conductivity ratio) into salinity units. The data were entered by a pair of operators (usually two student assistants) that would double-check the typed values to prevent data entry errors. The ‘autosal’ program prompted for the ambient and bath temperatures as well as the conductivity ratios of IAPSO Wormley seawater ampoules, substandards, and salinity samples, converting the conductivity units to salinity units using the bath temperature and the Autosal’s reading.  On July 18, 1991, Sean Kennan and Carl Chun revised the program to include the following: 1) to calculate the standardization offset from IAPSO measurements, 2) to incorporate this offset into subsequent calculations of salinity, 3) to make backups of data files during the run to avoid disastrous loss of data, and 4) to get time and date from the computer.  Originally, the program was on Personal Computers but was later moved to the Sun UNIX systems.  Additional improvements to the program were made by Sophia Asghar in November 1992, and June 1993.  Further updates and program improvements were made by Fernando Santiago-Mandujano in April 1995.   

 

On March 10, 2000, a few months after switching to using the Autosal 8400B for salinity measurements, an Autosal interface was purchased. The interface is used for direct downloads of conductivity data from the salinometer to a PC-based computer.  The program, which came with the interface (asal ), creates raw data files for subsequent data processing, as the salinity is measured by the Autosal.  The data are then transferred to the Sun UNIX system.  This interface was first used for HOT-114 salinity samples on May 2, 2000 (Figure 1).  The use of this interface eliminated the necessity for handwritten readings off the salinometer as well as manual data entry into a computer, thus increasing efficiency and decreasing operator errors.  For details on processing programs, see Section 2.3.1.  

 

1.2 Operators

 Table 2 below lists all Autosal operators, the model of the Autosal that they used, and the HOT cruises for which salinity samples were measured by each individual.

 

 

*Ted Walsh did not work for the Roger Lukas group but was given salt samples to measure on the Minisal.  Even after switching to 8400A, Sean Kennan had Ted Walsh measuring salinity samples for comparisons between Minisal and Autosal 8400A (Appendix B).

 

1.3 Maintenance

Proper maintenance is of extreme importance in acquiring reliable data from the Autosal. All maintenance activities must also be documented in detail.  There are two locations that must be kept up-to-date where maintenance documentation is concerned: the Autosal Run Log and the Maintenance Database on a PC.  In order to facilitate the entry of documentation, it is recommended to categorize maintenance events into two categories: regular maintenance and repair.

 

1.3.1 Regular maintenance

Each Autosal operator should keep their eyes open for signs of normal operation, such as flashing light bulbs, clean conductivity cell, smooth flow in and out, etc.  Cleaning the conductivity cell is an extremely important regular maintenance task that must not be ignored.  It is best to clean the cell every one or two cruises in order to attain the best readings from the Autosal.  Light bulbs within and associated with the Autosal must also be closely maintained, for the flashing lights control temperature, of which any fluctuation might greatly affect salinity data.  These light bulbs are regulated by thermistors, which control bath temperatures and will begin to fail after 5-7 years of use.  Thus it is recommended by Guildline to measure the bath temperature every two weeks to monitor thermistor conditions, if the Autosal is used constantly.  We also suggest changing both light bulbs in the case of failure of one bulb.  The failure of one indicates the likely failure of the other bulb shortly after.  Regular maintenance may also include changing substandard batches, changing bottle sets, replacing broken or damaged sampling bottles, and so on.  

 

All regular maintenance must be documented in the Autosal Run Log.  The Run Log binder contains template sheets for the Run Log, as well as the Salinity Measurement Sheets, which must be refilled and kept in stock when they run low.  An example Run Log is included in this report, listing all regular maintenance and some activities done on the Minisal or Autosal up until the completion of this report (Appendix J).   The activities preceded with an asterisk have reports included in respective appendices.

 

1.3.2 Repair

Repair cases involve the Electronics Technician to diagnose problems and possibly replace parts of the Autosal.  These events must be documented in the Run Log and in the Maintenance Database.  Table 3 lists each type of instrument, the HOT cruise numbers for which salinity samples were measured on the instrument, the Serial #, the UH Decal # given to the instrument, and the Index # assigned to the instrument in the Maintenance Database.  Keeping the database up-to-date is primarily the responsibility of the Electronics Technician; however, the Autosal operator who was present at the time of the maintenance event must also take responsibility in documenting the event in the Database.

 

 

Table 3: Salinometers used by HOT PO group
Instrument	HOT Cruise #	Serial #	UH Decal #	Index # in Database
AGE Minisal	1-26	n/a	n/a	n/a
8400	none	n/a	916RF	118
8400A	27-107	58296	065QV	283
8400B	108-present	63903	924SP	277

2. METHODOLOGY

2.1 Autosal Operation

Every Autosal operator must follow the procedures outlined in the most current version of the Autosal Measurement and Salinity Data Processing Guide (S. Asghar et al., July 2003 update) (Appendix K) for measuring salinity samples.  The Processing Guide has been revised numerous times in the past and its contents has changed significantly over the years.  Some procedures had only been passed on by the word of mouth and had never been documented until now.  Thus, it is extremely important to consult the up-to-date version of the Processing Guide for reference.  An appropriate copy must also always be placed as the first page in the Autosal Run Log binder next to the Autosal for quick reference.  Maintaining consistent procedures will help decrease operator errors and increase the quality of our data that we present to the scientific community.  The following emphasizes a few key points that every Autosal operator must keep in mind:

 

2.1.1 Duplicate measurements

Previous salinity reports and Autosal logs suggest that duplicate salinity samples were not always taken in order to cross-check the quality of the measurements made by a new or inexperienced Autosal operator.  Before HOT-33, duplicate samples were taken and measured in order to compare the performances of the Autosal to the Minisal.  From HOT-33 to present, duplicate salinity samples were taken from the first deep cast at Station ALOHA for cross-reference between primary and duplicate samples to check the quality of the Autosal operator’s measurements.  Duplicate samples are also important as they are back-up samples of the deep cast in  case of  problems during primary sample measuring and analysis. 

Duplicate samples should always be drawn from the rosette in conjunction with primary samples when sampling out at sea.  Duplicate samples for HOT-70 were saltier than the primary samples by as much as 1.6 mpsu, likely due to the fact that the duplicate samples were drawn from the rosette approximately 3-6 hours after primary samples were drawn. This type of incidents are preventable by following the regular procedure of drawing the duplicate sample from each Niskin bottle immediately after the primary.

 

After each HOT cruise, salinity samples should not be left for more than a week before measuring (except when there is limited time between consecutive cruises, or if there are problems with the Autosal). Evaporation and salt crystal formation in the sample bottles are both important factors that develop with time, and may significantly affect salinity values.  Duplicate samples should be measured at a date close to when primary samples are measured in order that they be measured in similar environments.  The best time to measure duplicate samples is after the primary salinity operator is done with all primary samples (one or two days in between is fine).  This way, if there are any problems in the measurement session that are indicated by substandard IAPSO values, the data are more easily corrected.

 

In the past, duplicate and sometimes even triplicate samples were taken from selected casts either for training purposes for a new Autosal operator (Kennan et al. 1992, “Summary of Training S. Asghar and R. Domokos on the Guildline Autosal,” Appendix L) or for comparison between glass and plastic bottles.  Also, due to insufficient number of glass bottles, plastic bottles were used for some duplicate samples before HOT-84 (see Section 1.1.2). Evaporation during the use of plastic bottles, especially if measured 15 days or more after the cruise, seemed to affect the data quite significantly.  Thus, triplicate samples were sometimes taken with some glass bottles for comparison. 

 

For HOT-70 to HOT-86, duplicate samples were taken from both deep casts.  For example, during HOT-78, duplicate samples were taken from both S2c1 and S2c12 (deep casts at Station ALOHA).  Molly Lucas measured the primary samples while Matt Cochran measured the duplicate samples for S2c1, and Matt Cochran measured the primary samples while Craig Nosse measured the duplicate samples for S2c12.  In this manner, the data allowed cross-checking of the quality of measurements by three Autosal operators.  In more common instances, such as for HOT-80 through HOT-86, duplicate samples were taken from two casts (e.g. S2c1 and S2c9), and while the primary Autosal operator measured primary samples from both casts, one duplicate operator measured duplicate samples from S2c1 and another duplicate operator measured duplicate samples from S2c9.  These methods were often employed to familiarize a new Autosal operator with measurement methods.  Sometimes, duplicate samples were taken from shallow casts such that the trainee would not affect primary salt measurements.  Also, in several isolated cases such as HOT-73, HOT-83, and HOT-149, triplicate samples were taken from one of the deep casts in order to have two duplicate operators for one cast.  This way, they were able to compare the data by the two duplicate operators to each other as well as with the data by the primary operator.  

 

2.1.2 Bath and ambient temperature range

Temperature fluctuations may greatly affect salinity measurements.  Accuracies of +/- 0.001 are possible on Guildline Autosal salinometers only when the measurements are made in conditions in which the temperature is maintained within 1-2^o C throughout the entire measurement run.  When ambient laboratory temperature fluctuates more than 1-2^o C during the analysis session, a considerable drift is noted (see Figure 1 in Stalcup, 1991, Appendix A).  As mentioned in the “Preparation” section of the Processing Guide (Appendix K), the temperature of the room in which the salt samples are stored and in which the measurements are to be made should be maintained at a range of 21-23^o C.  Bath temperature is usually fairly consistent and remains at 24^o C, but it is recommended to check the bath temperature as often as possible.  In the asal program, salinity units are converted from conductivity units assuming a bath temperature of 24^o C, and thus the bath temperature must be noted if it shows anything other than 24^o C. 

 

Ambient temperature is measured using a digital thermometer placed near the salinity sample boxes away from the Autosal.  This prevents the thermometer from being affected by the heat of the Autosal (from the thermistor and the light bulbs keeping the bath temperature constant at 24^oC) and allows the thermometer to measure the temperature of the area in which the samples have equilibrated.  Figure 2 is a graphical presentation of room temperature fluctuations during salinity measurement runs of previous HOT cruises.

Figure 2. Ambient temperature fluctuations during measurement runs for HOT cruises 22 through 149

 

Figure 2 shows the range of ambient temperature (maximum temperature - minimum temperature within a measurement run) plotted against HOT cruise numbers.  A red line is drawn at 2^o C as a threshold line for relatively stable data.  According to this criterion, data from HOT 22-80 are relatively stable except for two outliers at HOT-41 and HOT-47.  Data seem suspect from HOT 81-106, and from HOT 116-140. 

 

If temperature fluctuations do affect the salinity measurement values, variability could be detected by a drift or unusual changes in substandard measurements throughout the measurement run.  Table 4 shows the cruises marked suspect above the red line in Figure 2, and whether any adjustments were done to the respective data due to substandard variability.  Data were “corrected” if a drift was detected in substandard measurements throughout the salinity measurement run.  Data labeled “no corrections needed” indicates that no substandard drift was detected, and thus the data did not need any corrections.  For measurements during HOT-91, 96, and 128, variability was apparent in the substandard measurements but no corrections were made because the drift was insignificant.  For HOT-92 and HOT-101, the Autosal operators re-standardized the machine in the middle of the measurement runs and still corrected for a substandard drift after the measurement run.  For HOT-100 and HOT-106, the machine was also re-standardized in the middle of the measurement runs but no corrections were needed at the end of the run due to no drift in substandards detected. 

 

2.1.3 Good Autosal operator habits

In addition to controlling temperature fluctuations, every Autosal operator should maintain good measurement habits:

1. Make sure we have enough substandard water for this cruise and the next cruise.  If not, arrange to obtain more seawater during the upcoming cruise. 

2. The ZERO reference must be close to zero, and the STANDBY reading should remain stable. 

3. See that the lamps that regulate bath temperature are flashing on and off. 

4. Keep things clean.  Kim wipes must be used consistently and with care when measuring salinity samples. 

5. Clean the conductivity cell regularly.  After each session, rinse the cell thoroughly with distilled water.

6. Shake the bottles hard!!!  This will minimize the chance of having salt crystals that may affect the measurements.

7. Never adjust the standardization knob after the machine has been initially standardized for a measurement run.

8. Keep an eye out for abnormal substandard readings. They could be an indication that the Autosal might be experiencing problems during a measurement session.

9. Adjust the suppression knob when the reading is blinking or when indicated by the asal program.

10. Measure a substandard every 24 samples and when ending a session. Measure an IAPSO after the substandard if ending the last session or a duplicate session.

11. Keep detailed notes in the Measurement Sheets, the Run Log, and the Maintenance Database.

 

2.2 Documentation

The topic of documentation deserves a full section because this is the most important “good habit” an Autosal operator should maintain.  This history analysis is a review of only what has been left behind by documentation for all the years of measurements, and what has never been recorded in some form or another will never be reviewed or noted.  Thus, there may be information that we have not included in this report that we will never know due to poor documentation. 

 

2.2.1 Measurement Sheets / Run Log

During HOT-42 to HOT-114, the group used measurement sheets to manually record the measurements from Autosal 8400A.  Though this increased accuracy and organization in recording before, during, and after session information such as temperature, ZERO, STBY #, time, and standardization knob setting, manual recording of Autosal readings was messy and often times unreliable.  During HOT-114, the group started using the Autosal interface but still used the measurement sheets until the interface was working properly.  The use of measurement sheets stopped for HOT-115, and the use of the interface improved data capture accuracy while recording other facts began to be ignored.  It is difficult to find consistent notes regarding observations or temperature fluctuations during HOT-115 to HOT-140.  From HOT-141, we started using the measurement sheets again in order that we could record important ancillary information in an organized, consolidated manner.  The new Autosal Salinity Measurement Sheets can be found as Sal_Sheets.pdf, and should aid an operator in recording all pertinent information during a measurement run or session.  Completed Measurement Sheets, along with the printouts from the Autosal interface, should be filed into the Autosal Measurement Log binders. In addition, anything and everything done to, near, or concerning the Autosal must be recorded in the Autosal Run Log (sheets can be found as Run_Log.pdf) and filed into the Autosal Run Log binder.  We must get into the habit of recording everything in one consolidated area so that the next person looking for additional information need not look anywhere else and can feel confident about having all necessary information.

 

2.2.2 Written reports

In the past, written salinity reports have often included vague or incorrect information.   The common practice in writing Salinity Measurement Reports is to obtain a copy of the Report from a previous cruise and modify all relevant information relating to the current cruise.  However, in doing so, many operators have missed making minor changes, such as dates of the substandard batch, number of samples, etc.  It is also very important to note in the report any corrections that were done to the data, for the first place a person would go when he/she wishes to look back at salinity data is to the Salinity Reports.  It is always better to have more information written than no information.  It is highly recommended for the person writing the report to be meticulous in recording all corrections and unusual events in the report and to carefully check the Salinity Measurement Report before filing it away into the Salinity Data Processing Reports binder. Data corrections or modifications made after filing away the report must also be added to this binder.

 

2.3 Data Processing

2.3.1 Changes in processing programs

Prior to the Autosal 8400A, measurements made by Ted Walsh using the Minisal were also processed by him.  However, new processing parameters and methods were developed when the group purchased and started using the Autosal 8400A for HOT-27.  Currently, data from the Autosal measurement sessions are captured directly from the Autosal 8400B to the laptop hard drive using the automated data entry interface and the asal program that came with the interface.  The asal program creates raw data files (raw, *.lst, *.dat) which are necessary for data processing.  This method has been employed since May 2, 2000 (HOT-114), when the group started using the Autosal interface.  In addition, many original processing programs were modified to work with the asal program and its output files.  Besides processing programs, the most important difference between data processing before and after the interface is the method of data entry.  Before the interface, data acquisition and entry were all done manually: the Autosal operator recorded the conductivity ratios onto the Salinity Measurement Sheets, and student assistants computed the averages and entered them into the computer.  This method allowed for errors in data entry, caused by operator errors and confusion.  Detailed procedures are described in the Salinity Data Processing Guide (Pre-asal Procedures) (Asghar et al. 1994, Appendix M).

 

Table 5 is a summary of changes in processing programs that occurred after the purchase of the Autosal interface:

 

Table 5: Changes in Salinity Data Processing Programs
	Before Interface (Before HOT-114)	After Interface (HOT-114  to present)
Data acquisition	Manual entry on measurement log sheets, average salinity computed manually	Interface and the asal program used.
Data entry into a Sun computer hard drive	Manual entry into the computer, using autosal.c which outputs hot###a.log, hot###a.raw per session	Data files (raw, *.lst, *.dat) transferred through FTP, asal2dat.m reads the output of the asal program to create hot###.dat, ss###.dat, ss###.stat
Substandard statistics	ssstat.m	pssstat.m (to work with files created by asal2dat.m)
Application of corrections	raw2dat.c creates hot###.dat	hot###.dat already created by asal2dat.m during data entry
Autosal vs. CTD plots	plotdat.m outputs hot###.ps	autoplot.m (same as plotdat.m but this automatically creates plots sequentially w/out prompts)
Duplicates data	none	duplicates.m plots primary and secondary samples to determine outliers. duplicates2.m plots duplicate-primary salinity differences with CTD data.
Data transfer to CTD processing	dat2sal.c	dat2sal.m (same as dat2sal.c, just in Matlab)
Summary report	Word processing on Open Windows, tables created by *.trf and *.tbl (troff files)	Word processing on FrameMaker5, tables created by enscription.

2.3.2 Corrections

During the course of the HOT program, corrections have been applied to the salinity data if needed.  When salinity samples were being measured on the Minisal by Ted Walsh (HOT 1-26), the Minisal was not capable of standardizing by adjusting a standardization knob like an Autosal.  Thus, adjustments to the salinity data measured by the Minisal had to be determined by measuring an IAPSO standard seawater ampoule.  In this way, the difference between the measured IAPSO value and the expected IAPSO value was applied as correction to standardize the salinity data.

 

Other instances that required corrections to the salinity data were when 1) the IAPSO value at the end of a run differed from the previous IAPSO value by +1 mpsu or more, or 2) a drift or variability more than +1 mpsu was detected in the substandard measurements throughout the measurement run.  Both of these instances were an indication of drift in the Autosal measurements, and the data captured during that session or run were corrected accordingly.  Thus, it is important to keep an eye out for abnormal IAPSO or substandard values during a measurement session (see Section 2.1.3, #8), and if noted, the salinity supervisor must be consulted before taking further action.    

 

Corrections have also been applied in the past if conditions around the Autosal were not maintained.  These conditions have included failure of flashing light bulbs that control the thermistor to maintain constant bath temperature (for instance HOT-51 below) as well as large fluctuations in ambient temperature (see Section 2.1.2).  Evaporation corrections have been applied to samples collected in plastic bottles, if not measured within 15 days after the cruise (see Section 1.1.2).   

 

Sharon DeCarlo noticed that HOT-51’s second deep cast (S2c19) was noticeably saltier than almost 200 deep casts in history due to failure of the Autosal’s flashing light bulb (see Figure 1 in Sharon DeCarlo’s “HOT-51 Salinity measurement corrections,” Appendix N).  Details of the corrections are explained in Sharon DeCarlo’s report, but corrections to this cruise’s data done before DeCarlo’s corrections were difficult to find or if found, incomplete.  This emphasizes the importance of documentation, especially at times when no experienced person is present during measurements. All corrections done to the salinity data must be consolidated in the written Salinity Measurement Report.  Maintaining continuity of the HOT salinity history is only possible through clear and consistent documentation.

 

3. STANDARDIZATION

As a subset of the Methodology section, this section will discuss specific methods concerning IAPSO seawater ampoules and substandard seawater. 

 

3.1 IAPSO Standard Seawater

Wormley IAPSO seawater ampoules contain standard seawater with a “known” salinity value.  The label on the IAPSO vial shows the “Date” on which the IAPSO batch was made, the K_15, 2 x K₁₅, and the salinity value.  IAPSO ampoules are used as universal standards when measuring salinity samples.  Table 6 below shows the IAPSO batches used for HOT cruises 1-157.  HOT-21 (November 17-18, 1990, Chief Scientist C. D. Winn) is excluded from this table because the crane on R/V Na¢ina was too long for a safe deployment of the CTD and the cruise was aborted due to increasing seas and gale force winds.

 

3.1.1 Standardization methods with IAPSO standard seawater

IAPSO standard seawater has been used from the beginning of the HOT project.  When salinity samples were measured by Ted Walsh on the Minisal (HOT 1-26), IAPSO water was used to determine the correction value which was to be applied to the data for standardization (see Section 2.3.2).  When the group started using Guildline Autosal salinometers (8400A and 8400B), IAPSO ampoules were used before the measurement run to standardize the machine with the standardization knob and after the measurement run to check for drift in Autosal electronics.    

 

Standardization is the process of adjusting the Autosal electronics by turning the standardization knob until the Autosal measures the salinity of a sample with a “known” salinity value: an IAPSO sample.  Standardization is done at the beginning of a HOT salinity measurement run which usually lasts about 3-5 days.  The standardization knob is adjusted only if the observed reading of an IAPSO exceeds the expected reading by +/- 0.00003 conductivity units.  The amount of standardization knob adjustment (initial and final knob settings) and the Autosal conductivity readings associated with the knob adjustment at the time of standardization must be noted in the Salinity Measurement Sheets.  At the end of the session in which the last primary salinity samples are measured, another IAPSO is measured to check the Autosal for drift during the measurement sessions.  The standardization knob must NOT be adjusted after initial standardization.  Another IAPSO is measured not at the beginning of the duplicate session but at the END of the duplicate session.  This is also to check for Autosal drift during the duplicate session, and the standardization knob must NOT be adjusted. 

 

In the past, the process of initial standardization was incorrectly repeated during salinity measurement sessions when the Autosal drifted for various reasons such as sharp changes in room or bath temperature, leaks in the Autosal plumbing, unstable readings, or bad substandard measurements that may have suggested a drift in the Autosal.  In these events, an IAPSO was measured in order to determine the correction value for the set of data affected by the event, and the standardization knob was adjusted.  This was the case during HOT 67-106, when an IAPSO was measured and the standardization knob was adjusted at the beginning and at the end of each session and the duplicate session.  This way, the machine was standardized each time a session was started or when the machine seemed to drift more than the predetermined threshold value.  We do not recommend following this practice, because if a constant autosal drift existed during the measurement run, it would be corrected in steps marked by each standardization event.  The appropriate correction would consist of calculating the autosal drift from the substandard measurements, and then correcting each sample by this drift.  It is important to maintain consistent methods so that the drift of the Autosal electronics can be detected in a uniform context. 

 

We recommend following the proper procedures of standardization using IAPSO seawater ampoules concerning initial standardization, standardization knob adjustment, and IAPSO frequency.

 

3.1.2 Autosal electronics drift

 Though the frequency in which the standardization knob was adjusted has not been consistent throughout the years (see previous Section 3.1.1), it is possible to determine the drift of the Autosal electronics throughout the HOT salinity history.  By plotting the changes in standardization knob setting (SKS) against the changes in salinity values shown on the Autosal at the time of standardization, the equivalent change in salinity units per knob setting units for Autosals 8400A and 8400B were determined (Figure 3). 

 

Figure 3. Changes in the Autosal’s standarization knob setting and their corresponding change in salinity, for cruises 33 through 80 (a), 81 through 107 (b), and 108 through 148 (c).

 

The plots are divided into three categories: HOT 33-80 (Fig. 3a), HOT 81-107 (Fig. 3b), and HOT 108-148 (Fig. 3c).  HOT-33 is the first cruise in which changes in SKS along with changes in salinity units were noted in the salinity measurement sheets.  During HOT-80, the Autosal 8400A had to be readjusted because the standardization knob reached its minimum setting (0/00) and could not turn any further (see Section 1.1.1).  After the knob was readjusted, the group resumed using the Autosal 8400A for HOT-81 but started standardizing more than once during a measurement run (see Section 3.1.1 above).  This continued until HOT-108, when the group started using the Autosal 8400B and resumed practice of standardizing the machine only during initial standardization.  Regardless of the readjustment and the change in standardization methods, the standardization knob adjustments for the Autosal 8400A (from HOT 33-107) equaled 1.42 x 10^-4 psu/knob setting unit (Figures 3A and 3B).  For the Autosal 8400B, the standardization knob adjustments equaled 2.04 x 10-4 psu/knob setting unit (Figure 3C). 

From these calculations, drift of the Autosal electronics was calculated (Figure 4).

 

Figure 4. Electronics drift for Autosals 8400A (a), and 8400B (b) during HOT cruises 33 through 148.

 

Figures 4A and 4B show the equivalent change in salinity each time the SKS was adjusted in Autosals 8400A and 8400B, plotted over time.  Each data point was recovered from Autosal log sheets and represents the equivalent change in salinity at time of one of the following: SKS change at the beginning of the salinity session, SKS change after initial standardization, SKS change if the knob was adjusted during the salinity session, or SKS change from standardization at the beginning of the duplicate session.  The solid red lines indicate the drift rates estimated for each category. Apparently, when Autosal 4800A standarization knob was adjusted (after cruise 80), something else was adjusted that reduced its drift rate. However, there is no documentation describing what type of adjustment was made to the Autosal. The maximum drift in the Autosal electronics throughout history occurred during HOT 33-80 on the Autosal 8400A, at -5.3426 x 10^-5 psu/day (Figure 4A, left).  Assuming a maximum measurement run of five days, the maximum drift would be approximately 0.27 mpsu during the salinity measurement run.  As this drift is very small, we can conclude that Autosal electronics drift has not affected significantly the HOT salinity measurements.  

 

3.1.3 IAPSO batch comparisons

IAPSO batch-to-batch comparisons are necessary to determine whether IAPSO differences between batches may affect salinity data over time.  In October 1995, duplicate salinity samples from the BATS (Bermuda Atlantic Time-Series) project were measured for comparison with HOT-67 salinity samples.  The mean difference between the HOT-BATS duplicate salinity samples was 1.3 mpsu (Nosse 1996, “HOT-BATS Salinity Intercomparison,” Appendix O).  Nosse speculated that the only differences in these measurements were that IAPSO batch P127 was used by BATS for their measurement sessions, while IAPSO batch P123 was used by the HOT group for their measurement sessions.  However, a recent study of IAPSO water comparison by Aoyama et al. (2002, Appendix P) stated that the difference between batches P123 and P127 is only 0.1 mpsu. Aoyama et al. (2002, Appendix P) suggests that within-batch comparisons are useful in indicating the presence of drift in Autosal electronics. Their study also determined correction values for IAPSO batches up to P129.  Their batch-to-batch comparisons indicate an IAPSO shelf life of 5 years.  In the future, we should continue monitoring IAPSO within-batch differences to detect possible Autosal drift.  The corrections to IAPSO bottles suggested by Aoyama et al. (2002, Appendix P) should be applied to the HOT samples.  

 

3.2 Substandard Seawater

 

Since HOT-24, samples of secondary lab-standard (substandard) seawater have been measured periodically during the measurement run to monitor possible drifts in Autosal electronics.  The use of substandard seawater saves the expense of numerous costly IAPSO standards.  A substandard seawater batch is a 50-liter glass carboy filled with seawater and topped with two inches of mineral oil, which acts as an evaporation barrier.  Substandard batches are made from 60 liters of seawater collected from 1000 m at Station ALOHA during a HOT cruise.  The mean salinity of the substandard batch is determined by multiple comparisons with IAPSO standards over the substandard batch lifetime. The idea is to have a large, stable batch of seawater of known salinity with which measurements using the autosal can be monitored.  Table 7 below shows the substandard batches used for each HOT cruise and the date each batch was made.

 

 

 

*Batch #7a was made as an emergency substandard for HOT-52 out of the remaining seawater used to make Batch #7.  The salinity of this batch was higher due to evaporation in the plastic carboy over time (Figure 5a).

**Batches #11 and #13 had lower salinities than other batches because of mixed seawater from depths other than 1000 m (see Section 3.2).

 

3.2.1 Substandard history

Salinity samples from HOT cruises are generally measured over a 3-5 day period during the week after the cruise.  Prior to starting a measurement run, the salinometer is standardized using IAPSO standard seawater ampoules. After standardization, two or more substandard seawater samples are measured until a value is repeated within 0.00002 twice-the-conductivity-ratio units from the previous measurement. The result serves as a baseline against which to compare subsequent measurements of the substandard over the period during which the salinity samples are measured. The salinity of the substandard does not change significantly over the course of a week (see Kennan, 1992 analysis below), so that changes in its measured salinity from day to day represent drift or jumps in the Autosal's electronics, which may need to be corrected.  Corrections to the Autosal measurements are usually applied when the substandards differ from the substandard session’s mean by more than +/-1 mpsu.

 

In his “Summary of Lab-Standard Salinity Measurements from 12/27/90 to 8/11/92”, Sean Kennan (1992, Appendix R) compared mean substandard salinities from substandard batches #1-5.  He concluded that the standard deviation of substandard measurements during a session was always less than 1 mpsu, and after July 1991, less than or equal to 0.5 mpsu.  His results showed that substandard water was indeed adequate to monitor drift of the Autosal during our measurement sessions.  His calculations of substandard drift over time showed that substandard water has an insignificant drift of –0.4 ± 0.8 mpsu over 100 days. He attibuted the 0.8 mpsu scatter to variations in the Autosal’s electronics, or to differences between IAPSO standards.

 

Figure 5 shows the mean salinities of substandard seawater batches for each salinity measurement run over time (one run per cruise). A line connects mean salinities of the same batch, and error bars (in Figure 5B) indicate the standard deviation of each substandard run. 

 

 

Figure 5. Mean salinities of the Substandard batches 1 through 31 used during the HOT project (a), and after removing batches 1-3, 7a, 11, and 13 (b). The circles indicate the mean of each salinity measurement run, and the bars in (b) are plus minus one standard deviation of the mean. Solid lines connect mean salinities from the same batch.

 

Most of the substandard batches have similar salinities between 34.4 and 34.5.  Seawater used for batches #1-3, #7a, #11, and #13 had mean salinity values that were very different from the rest of the substandard batches.  For substandard batches #1-3, mercuric chloride was used to prevent biological growth and as it precipitated out of the seawater (as mentioned above), the salinity of the substandard changed (Kennan 1992, Appendix R).  Batch #7a was made as an emergency substandard for HOT-52 because batch #7 was running low.  Batch #7a was prepared from the remaining seawater used to make batch #7, and thus had a higher salinity due to evaporation in the plastic carboy over time.  The cause of lower salinities for batches #11 and #13 are likely due to mixed seawater from depths other than 1000 m.  Batch #11 had a lower salinity than other batches because 10 of the 60 liters used to make the batch came from 500 m (salinity minimum) instead of 1000 m.  Batch #13 also had a lower salinity than other batches.  Batch #13 was collected during HOT-76, but only five Niskin bottles (12 L) were fired at 1000 m, providing an insufficient amount of seawater for the substandard batch.  We hypothesize that batch #13’s low salinity is due to mixing seawater collected from a different depth, but there is no documentation regarding the preparation for this substandard batch. 

 

Batches #1-3, #7a, #11, and #13 have been removed as outliers in Figure 5B, which shows a closer look at the range of mean substandard salinity variability.  The range of the mean substandard salinity variability is about 0.04 psu. For comparison, Figure 6 shows the mean CTD salinity variability at Station ALOHA from HOT 1-140 at 1000 dbar, the depth from which substandard water is sampled.

 

Figure 6. Mean CTD salinity at 1000 dbar during HOT cruises 1 through 140.

 

Although most of the substandard batches’ salinities correspond to the salinities in Figure 6, many of them have higher salinties than those observed at 1000 dbar (e.g. batches 5, 6, 9, 10, 20). This discrepancy may be due to evaporation of the sample in the plastic carbuoy before preparing the substandard, or due to water of a higher salinity level being included in the substandard. However, no documentation was found in this regard.

 

Though the salinity of the substandard does not change significantly during the course of a salinity measurement session, the mean salinity of substandard seawater from the same batch sometimes changed between sessions. For many of the batches, the salinity tends to decrease towards the end of the substandard batch lifetime.  We hypothesized that because saltier water tends to settle on the bottom of the carboy and the water is drawn out from the bottom, what remains at the end of the substandard batch lifetime is fresher water that settled higher up at the beginning of the substandard batch lifetime.  Some of the error bars are very large for some of the cruises, which could be because bad substandard measurements were included in the data set.  We suggest looking carefully at the data with large error bars to eliminate outliers before using these data for further analysis.

 

Biological growth in substandard water was noted in batch #20 in the HOT-108 Salinity Measurement Report (see Don Wright’s update report, Appendix S).  Biological growth did not seem to interfere with the substandard salinity during HOT-108 measurements.  There was concern for HOT-109, however, and thus the group decided to take nine samples of the affected substandard and store them away from light.  They also filtered the remaining substandard with a 142 mm Gelman 0.2 um pore size membrane filter, drew six samples and stored them away from light.  A comparison between the filtered and unfiltered samples showed a difference of approximately 1 mpsu higher for unfiltered samples.  It is important to learn from this experiment the gravity of keeping the substandard batch covered with dark plastic bags and away from light, for biological growth may affect the adequacy of the substandard water as a means to monitor drift of the Autosal.

3.2.2 Frequency of substandard measurement

The number of HOT cruises that can be measured with one substandard batch varies depending on the frequency of substandard measurement during a run. The frequency of measuring a substandard has varied since they were first measured during the HOT-24 salinity session. Initially, substandard measurements were used for comparison between Minisal and Autosal 8400A.  Starting with HOT-27, substandards were measured about every 24 salinity samples on Autosal 8400A and continued until HOT-38.  From HOT-39 onward, substandards were measured to provide a baseline measurement in case the Autosal drifted. 

 

The frequency of substandard measurements varied with the Autosal operator, and thus determined the life span of the substandard batch.  While most operators measured a substandard every 15-30 samples, the extreme cases have been a maximum at every 152 samples, and a minimum at every 6 samples.  While substandard batches #6-9 lasted for about six measurement runs, batches #26-28 were depleted quickly, after only two or three measurement runs (see Figure 5B).  We recommend measuring a substandard after every sample case (which holds 24 salinity sample bottles).  Since we draw water samples from all 24 Niskin bottles during the first couple casts for HOT cruises, we eliminate confusion and are able to maintain consistent substandard batch life spans.

 

When looking back at suspicious data, knowing operator habits may help determine causes for systematic problems, errors, or differences in the salinity data from specific cruises (see Section 1.2). 

 

4. QUALITY CONTROL

Preliminary screening of the water sample salinities is done by comparing against all previous data deemed reliable that had been collected at the particular site (Stations Kahe, ALOHA, Kaena, etc.). The nominally calibrated CTD salinity trace is also used to identify questionable discrete samples. Potential rosette mistrip problems are resolved, where possible, before data are excluded from use in the calibration of the conductivity cell.

 

After an initial calibration of the conductivity cell using all casts within a cruise, the deviations between CTD salinity and bottle salinity are tested against limits within 4 pressure ranges (see below). Bottle salinities outside these limits are marked as “suspect” or “bad”, and are not used in further iterations of the calibration.

 

The procedure for screening the bottles against calibrated CTD salinities includes a method to prevent the flagging of bottles with moderate CTD minus bottle salinity differences due to bottle closure and mark pressure mismatch, especially in high gradient regions. The method includes the calculation of the local vertical salinity gradient from a least squares fit to 5 levels of the 2-dbar averaged data on both sides of the mark. The absolute value of the CTD minus bottle salinity difference (ΔS), divided by the absolute value of the gradient yields a pressure Δp that is used as an additional parameter to determine the bottle quality.

 

The quality of each bottle is determined by comparing ΔS against the standard deviation (SD) of the ensemble of “good data”, and comparing Δp against a pressure cutoff value (P) as follows:

 

• Bottle is bad if              ΔS > 4*SD and Δp > P
• Bottle is suspect if         ΔS ≥ 4*SD and Δp ≤ P, or if 3*SD < ΔS < 4*SD and Δp > P
• Bottle is good if            ΔS > 3*SD and Δp ≤ P, or if ΔS ≤ 3*SD

 

The pressure cutoff value P was set to 7 dbar. This cutoff value represents the maximum vertical distance that a bottle can travel between the time of the mark and the closing of the bottle as a result of a ship roll or changing wire angle. A greater P value would imply vertical motion of more than 7 dbar, which is unlikely.

 

The salinity standard deviation values of the ensemble of “good” data (SD) used for the bottle screening in the 4 pressure ranges were: 0.0035 (0-150 dbar), 0.0048 (150-500 dbar), 0.0020 (500-1050 dbar), and 0.0011 (1050-5000 dbar).

 

In order to show the salinity sample variability in deep and bottom waters at the ALOHA station, bottle salinities from HOT cruises at levels below 3000 dbar are plotted in potential temperature coordinates in figures 7 through 11. Only bottles flagged “good” are included in the plots.

 

The salinity variability below 3000 dbar at the ALOHA station can also be seen in figure 12. This figure shows contours of bottle-calibrated CTD data, plotted versus potential temperature throughout the HOT program. Also indicated on the plot are the IAPSO standards (Table 6), and lab substandard batches used (Table 7), as well as the initials of the operator that measured the salinities (Table 2).

Figure 7. Bottle salinities below 3000 dbar for HOT cruises during 1988-1989, 1990 and 1991.

 

 

Figure 8. Bottle salinities below 3000 dbar for HOT cruises during 1992, 1993 and 1994.

 

 

Figure 9. Bottle salinities below 3000 dbar for HOT cruises during 1995, 1996 and 1997.

 

Figure 10. Bottle salinities below 3000 dbar for HOT cruises during 1998, 1999 and 2000.

 

 

 

Figure 11. Bottle salinities below 3000 dbar for HOT cruises during 2001, 2002 and 2003.

 

Figure 12. CTD salinity contours at Station ALOHA below 3000 dbar throughout the HOT cruises. The vertical dashed lines indicate the first time that the IAPSO batch shown in the upper axis was used (Table 6). Lab substandard batches used during cruises (Table 7) are indicated by the tick mark along the upper axis. The salinity operator measuring the samples (Table 2) is indicated below the lower axis (TW: Ted Walsh, SK: Sean Kennan, S-R-J: Sophia Asghar, Reka Domokos, Jim Potemra; JY: Jinchun Yuan, CN: Craig Nosse, ML: Molly Lucas, MC: Matt Cochran, DW: Don Wright, JJ: Jeremiah Johnson, NL: Noel Larson, DF: Daniel Fitzgerald, MI: Maya Iriondo). The onset of “cold events” (Lukas et al., 2001), is indicated by white vertical dashed lines. The tick marks along the lower axis indicate HOT cruises. The blue line indicates the mean temperature at the level of the sill depth separating the Maui sill from the Kauai sill (4450-4500 m).

5. CONCLUSION / SUGGESTIONS

We have reviewed the HOT bottle salinity history through its instruments, methods, standardization, and quality control.  The following list includes a summary of  suggestions to maintain and improve the data quality.

 

1.  [Section 1.1.2] An evaporation correction of 1 mpsu/14 days should be applied to data obtained from samples in plastic bottles that were measured on or later than the 15th day after the cruise.

2.  [Section 1.3.1] Each Autosal operator should keep their eyes open for signs of normal Autosal operation, such as flashing light bulbs, clean conductivity cell, smooth flow in and out, etc. 

3.  [Section 1.3] All regular maintenance must be documented in the Autosal Run Log.  The Run Log binder should contains template sheets for the Run Log as well as the Salinity Measurement Sheets, which must be refilled and kept in stock when they run low.  All repairs must be documented in the Run Log as well as in the Access 2000 Maintenance History Database. 

4.  [Section 2.1] Every Autosal operator must read the most current version of the Autosal Measurement and Salinity Data Processing Guide (Appendix K) before measuring any salinity samples. 

5.  [Section 2.1.1] Duplicate samples should always be drawn from the rosette in conjunction with primary samples when sampling out at sea.  After each HOT cruise, salinity samples should not be left for more than a week before measuring.

6.  [Section 2.1.2] Keeping constant bath and ambient temperatures when running salinity samples is crucial.  The bath temperature must be noted if the thermometer shows anything other than 24^o C.  The ambient temperature must be maintained within 1-2^o C of 21^o C. 

7.  [Section 2.2.2] The person writing the Salinity Report must be meticulous in recording all corrections and unusual events in the report and to carefully double-check the report before filing it away into the Salinity Data Processing Reports binder.

8.  [Section 2.3.2] All corrections applied to the salinity data must be consolidated in the written Salinity Measurement Report.

9.  [Section 3.1.1] Standardization methods using IAPSO seawater must be consistent.  The amount of standardization knob adjustment (if adjusted) and the associated conductivity readings must be written down in the measurement sheets.  The standardization knob must NOT be adjusted after initial standardization.  Furthermore, after initial standardization, two more IAPSO ampoules should be measured, one at the end of the measurement run and the other at the end of the duplicate session.    

10.  [Section 3.1.3] We should continue monitoring IAPSO within-batch differences to detect possible Autosal drift.  The corrections to IAPSO bottles suggested by Aoyama et al. (2002, Appendix P) should be applied to the HOT samples. 

11.  [Section 3.2.1] We suggest looking closely at substandard batch mean data for cruises with large error bars (Figure 5B) to eliminate outliers before further analysis.

6. REFERENCES: Appendices

A: Stalcup, M.C.1991. Salinity Measurements. WHP Operations and Methods - July 1991. Woods Hold Oceanographic Institution. 1-9.

B: Kennan, S. 1992. Summary of Comparisons between Guildline Autosal and AGE Minisal. 1-18.

C: Fitzgerald, Daniel S. 2002. Autosal 8400A and 8400B comparison.

D: Kennan, S. 1991. Evaporation in Plastic Bottles. 1.

E: Kennan, S. 1991. Summary of Salinity Measurements from HOT-23 (2/1-2/6,1991). 1.

F: DeCarlo, S. 1992. HOT Deep Salinity Data Summary. 1-3.

G: Kennan, S. 1991. How much mineral oil needed for carbuoy?

H: Valenciano, M. and Y. Rii. 2003. Autosal History Maintenance Database Report. 1-6.

I: Program printout: autosal.c. August 11, 1995.

J: Rii, Shimi. 2003. Autosal Run Log.

K: Asghar, S., C. Nosse, D. Wright, J. Johnson, N. Larson, D. Fitzgerald, S. Rii. 2003. Autosal Measurement and Salinity Data Processing Guide.  1-21. 

L: Kennan, S., S. Asghar, R. Domokos. 1992. Summary of Training S. Asghar and R. Domokos on the Guildline Autosal. 1.

M: Asghar, S., C. Nosse, D. Wright, S. DeCarlo. 1994. Salinity Data Processing Guide (pre-ASAL procedures). 1-7.

N: DeCarlo, S. 2003. HOT-51 Salinity measurement corrections. 1-4.

O: Nosse, C. 1996. HOT-BATS Salinity Intercomparison. 1-2.

P: Aoyama, M., T.M. Joyce, T. Kawano, Y. Takatsuki. 2002. Standard seawater comparison up to P129. Deep-Sea Research I 49: 1103-14.

Q: Nosse, C. 1997. Cross-calibration of Standard Seawater Batches P114, P115, P118, P121, P123, P128 and P130 through the Hawaii Ocean Time-series. 1-14.

R: Kennan, S. 1992. Summary of Lab-Standard Salinity Measurements from 12/27/90 to 8/11/92. 1-4.

S: Wright, Don. 1999. Update report on substandard biological growth experiment.

Lukas, R., F. Santiago-Mandujano, F. Bingham, and A. Mantyla, 2001. Cold bottom water events observed in the Hawaii Ocean Time-series: implications for vertical mixing. Deep-Sea Research I, 48 (995-1021).