|
|
|
Funded by the National Oceanic and Atmospheric Administration's
Environmental Services Data and Information Management Program
(ESDIM).
The five RITS ground-based in situ monitoring sites:
Barrow, Alaska;
Niwot Ridge, Colorado;
Mauna Loa, Hawaii;
Cape Matatula, American Samoa;
South Pole, Antarctica.
The Climate Monitoring and Diagnostics Laboratory (CMDL) of the National Oceanic and Atmospheric Administration (NOAA) operates four baseline observatories at Barrow, Alaska (71.3o N), Mauna Loa, Hawaii (19.5 o N), Cape Matatula, American Samoa (14.2o S), and South Pole, Antarctica (90.0o S). A fifth site at Niwot Ridge, Colorado (40.4o N) is run cooperatively with the University of Colorado. Since 1977, pairs of air samples collected weekly in flasks at these sites have been transported to Boulder and analyzed by CMDL's HATS (Halocarbons And other Trace Species) group for several compounds known to play an active role in the chemistry of stratospheric ozone.
In 1985, the Radiatively Important Trace Species (RITS) program was initiated to provide concurrent in situ monitoring of several of the same ozone-depleting chemicals measured in the flask program. Gas chromatography (GC) systems with electron capture detectors (ECD) were installed at the five sites listed above for measuring the chlorofluorocarbons CFC-11 and CFC-12, the chlorinated solvents methyl chloroform (CH3CCl3) and carbon tetrachloride (CCl4), and nitrous oxide (N2O). Secondary calibration standards referenced to primary gravimetric standards were prepared in the laboratory and shipped to the sites. Initially, samples were injected once every hour at the sites with environmental sampling occurring once every three hours (i.e. two calibration samples and an environmental sample every three hours). In 1991, the systems were modified for sampling every half hour. Calibration and environmental samples were alternated to increase the environmental sampling frequency from three hours to one hour.
Since the establishment of these programs, many new compounds have been added to the list of those measured and analyzed by the HATS group. These include several hydrochlorofluorocarbons (HCFCs) and hydrofluorocarbons (HFCs) that have been developed to replace the CFCs banned by the United Nations' Montreal Protocol. Another compound that has gained considerable attention in recent years is Sulphur Hexafluoride (SF6) due primarily to its greenhouse potency (25,000 times more powerful than CO2) and longevity in the atmosphere (atmospheric lifetime > 3000 years). In response, the flask sampling program now analyzes over 25 gases, up more than eight-fold from the original three (i.e. CFC-11, CFC-12, and N2O), and the in situ monitoring program is upgrading to the new 4-channel CATS system GCs.
CATS systems are now in place at all five of the RITS field sites. As of October 2000 the RITS system at Barrow has been completely shut down, while the systems at Niwot Ridge and South Pole continue to run all available channels. At Mauna Loa and Cape Matatula, two of the three available RITS channels have been turned off. Those parts of the RITS program still in operation remain so because of unreconciled discontinuities between the RITS and CATS measurements. In order to insure the best possible transition from RITS to CATS, the RITS systems will not be shut down completely until all such issues have been resolved. Our current aim is to have the RITS program offline by the time South Pole closes for winter in February 2001.
During this transition, an effort is underway to bring the RITS data -- which now spans a period of almost 15 years -- into a standard form for inclusion in the NOAA data center archives. One of the goals of this effort is to produce a comprehensive and finalized set of atmospheric mixing ratios for the five compounds (listed above) that the RITS systems were originally designed to measure. In addition, there may be considerable useful information concerning newer compounds of interest that has yet to be extracted from the RITS raw data. In particular, small SF6 peaks have been evident in the third channel of the RITS system for several years. Thus, the second major goal of this archiving effort is to rebuild the RITS database in an easily expandable form that can readily accomodate future analyses of additional compounds.
The RITS in situ monitoring system is a 2 or 3-channel GC/ECD. The first channel -- or A channel as it is known locally -- uses a Porasil B separation column and P5 carrier gas (Ar/CH4 mixture) for the elution of N2O, CFC-12 and CFC-11 in chronological order. The second (or B) channel uses an OV101 clone separation column and N2 carrier gas for the elution of CFC-11, CH3CCl3 and CCl4. Channels A and B are associated with a 2-channel Hewlett-Packard 5890 GC. Finally, the third (or C) channel (not installed at South Pole) uses a Porapak Q separation column, P5 carrier gas and a single-channel Shimadzu GC. It is dedicated solely to the measurement of N2O. A simple schematic of the RITS system along with a description of the analytical technique and installation history at the four baseline observatories can be found at the in situ monitoring page. A photograph of the RITS system at Niwot Ridge is shown below.
The RITS system at Niwot Ridge. Shown from left to right are the
data-recording system, analog-to-digital converter boxes, Hewlett-Packard GC
(housing channels A and B), and the Shimadzu GC (housing channel C). On the
wall above the system is the environmental stream intake manifold. Cylinders
holding calibration gases, P5 and N2 carrier gases, and pnuematic
air (for driving the stream select valve) are out of the picture to the right.
A 3-channel RITS system sampling once every half hour produces 1008 chromatograms in a week. These chromatograms are recorded on site using an integer-differencing scheme and an internal (i.e. binary) proprietary format employed by the Hewlett-Packard machines that were chosen in the mid-1980s for recording and manipulating the data. Transport of raw chromatograms to Boulder has traditionally been accomplished through the mail on floppy disk, although more recently in some cases -– and in all cases with the new CATS system -– raw chromatograms have been retrieved over the internet. The last time the RITS program was fully operational was early in 1999. At that time ~4700 chromatograms per week were arriving in Boulder for processing. As of October 2000, the influx of RITS output has been cut to roughly one-half of its maximum.
The processing of chromatograms is performed on a Hewlett-Packard workstation using modified BASIC language software aquired when the RITS program was still under development. Binary format database files were designed in-house for storing the output generated by the chromatogram processing software for later retrieval during the computation of mixing ratios. Every database file is built around one of several database record structures each conforming to a particular multi-sample sequencing protocol that has changed over the years. The volume of incoming data and the physical constraints and cost factors associated with the computer data systems technology of the mid-1980s made necessary the efforts taken to economize the amount of space required for data storage and the time required for processing.
Currently, the RITS raw data archive consists of several million chromatograms stored on a total of 17 magneto optical disks with secondary backup to DC600 tape cartridges and zip archives stored on other hard drives. Over the last several months, well over half of the total raw data archive has been standardized and transfered to CDROM. It is expected that the remainder will be likewise standardized and transfered to CDROM by the end of year 2000. The database of processed chromatogram peaks is maintained on HP-UNIX workstations and backed up on a separate UNIX server as are the calibration mixing ratios for every calibration tank used in the RITS program.
Using the current method, it takes one person ~3 hours to reprocess a full week's worth of trouble free RITS data from the five sites. When problems occur that slow down processing -- problems requiring extra user attention and input -- the time required can increase dramatically. Thus, for a single person to reprocess all RITS chromatograms generated during the course of just one year would require a minimum of four 40-hour work weeks or on the order of an entire year to go back through all of the data in the archive. Troublesome data only adds to the amount of time required.
Given this situation, initial emphasis is being placed on generalizing and modernizing the database and raw data archive. This effort is motivated not only by the desire to save the raw data and facilitate further analyses and data recovery in the future, but also by the need to recover processed data that has been "misplaced" due to minor flaws in the current reduction process. A detailed look at the database has turned up three primary types of data misplacement during processing:
Most of these various types of data misplacement require the reprocessing of chromatograms to correct. (Up to now the affected data have tended to be excluded as outliers during the computation of mixing ratios.) In some cases, the chromatograms themselves have been permanently lost. For example, prior to 1994 logistical constraints peculiar to South Pole lead us to process chromatograms on site and save only the processed data. In other cases, reprocessing of chromatograms will not help without first restructuring the database so that each sample injection is uniquely timestamped. (See, for example, the third bulleted item above.) Restructuring is also required to make room for the addition of new peaks (e.g. SF6) that have yet to be analyzed.
The plan for finalizing the RITS data archive is outlined as follows:
A full-time CIRES employee, David Nance, was hired in June of 1999 and dedicated to the task of finalizing the RITS dataset. After a period of training during which he familiarized himself with the RITS systems and data processing procedures through the processing of a continuing influx of RITS data, he began an investigation of RITS database issues. He used what he learned from his investigation to develop a plan of action (outlined above) and a strategy for developing the software he would use to accomplish his tasks. He began the process of standardizing the 2 million+ chromatograms and saving them to CDROM in the spring of 2000.
This effort has been complicated by the variety of raw data file-naming and storage conventions that have resulted over the years from changes in RITS system hardware and software. It has also been complicated by the unforseen frailties of the previously chosen primary storage media for the raw data, the magneto optical disk. Though convenient when functional, we have at times been forced to go to much less convenient secondary tape backups to retrieve raw data files that we could not access on corrupted optical disks. We have also been forced on one occasion to do a block-by-block extraction and reassembly of raw data files stored on an optical disk with an irrevocably damaged file system in order to retrieve files that were overlooked during secondary backups to tape.
Apart from issues related to data storage, we have also found previously undiscovered consistency problems with a relatively small portion of the older data. Each chromatogram is tagged both internally and externally (i.e. in the filename) with information regarding the time at which it was generated and the RITS system channel that produced it. These internal and external descriptors are compared and occasionally disagreements are found that must be resolved. "Time folds" -- situations where unique chromatograms have overlapping time stamps -- can occur when an unlogged system clock change occurs at the field site. This leads to overwriting during processing if not detected and fixed. Sometimes it can take a fair bit of investigating to figure out what happened and what needs to be done to correct the problem and save the data.
David continues to work his way back through the raw data, checking it for consistency, resolving conflicts and reprocessing the corrected data as he goes, then saving the standardized files to CDROM. As of October 2000, he is approximately 75% of the way completed with this task and expects to be finished by the end of the year. Concurrently, he is working on the software required for building a restructured database, which he hopes to have implemented almost immediately upon completion of the chromatogram inventory.
Last modified October 2000 by David Nance.
 |
|