[an error occurred while processing this directive]
Total ozone trends at sixteen NOAA/CMDL and cooperative
Dobson spectrophotometer observatories during 1979-1996

W.D. Komhyr, G.C. Reinsel, R.D. Evans, D.M. Quincy, R.D. Grass, and R.K. Leonard

Reprinted from


Volume 24, Number 24, December 15, 1997

(Received August 19, 1997; revised November 3, 1997; accepted November 11, 1997.)
Copyright 1997 by the American Geophysical Union. Further electronic distribution is not allowed.


Total ozone trends at sixteen NOAA/CMDL and cooperative
Dobson spectrophotometer observatories during 1979-1996

W.D. Komhyr1,5, G.C. Reinsel2, R.D. Evans3, D.M. Quincy3, R.D. Grass5, and R.K. Leonard4,5

Abstract.    Ozone trends, derived from 1979-1996 Dobson spectrophotometer total ozone data obtained at five U.S. mainland midlatitude stations, averaged -3.4, -4.9, -2.6. -1.9, and -3.3%/decade for winter, spring, summer, and autumn months, and on an annual basis, respectively. At the lower latitude stations of Mauna Loa and Samoa, corresponding-period annual ozone trends were -0.4 and -1.3%/decade, respectively, while at Huancayo, Peru, the 1979-1991 annual trend was -0.9%/decade. A linear trend approximation to ozone changes that occurred since 1978 during austral daylight times at Amundsen-Scott (South Pole) station, Antarctica, yielded a value of -12%/decade. By combining 1979-1996 annual trend data for three U.S. mainland stations with trends for the sites derived from 1963-1978 data, it is estimated that the ozone decrease at U.S. midlatitudes through 1996, relative to ozone present in the mid-1960s, was -6.7%. Similar analyses incorporating South Pole data obtained since 1963 yielded an ozone change at South Pole (daylight observations) through 1996 of approximately -25%. South Pole October total ozone values in 1996 were lower than mid-1960s October ozone values by a factor of two. Trend data are also presented for several shorter record period stations, including the foreign cooperative stations of Haute Provence, France; Lauder, New Zealand; and Perth, Australia.


    Realization in recent years of the potential of chlorine released from man-made chlorofluorocarbons (CFCs) to destroy atmospheric ozone by homogeneous and heterogeneous photochemical processes (for a review, see Solomon [1990]), and measurement evidence of dramatic ozone destruction in Antarctica and to a lesser extent at midlatitudes, led to universal concern about possible ozone depletion on a global scale and the political formation in 1987 of the Montreal Protocol for the purpose of limiting the world-wide use of CFCs. In accordance with requirements of the Montreal Protocol, the scientific community has since prepared several World Meteorological Organization/United Nations Environment Programme (WMO/UNEP) ozone assessment reports that assess our knowledge of ozone in the atmosphere, with a focus on ozone depletion. Additional ozone trend analyses have appeared in the literature [e.g., Reinsel et al., 1997, 1994; Bojkov et al., 1995; McPeters et al., 1996; Harris et al., 1997]. An updated WMO/UNEP ozone assessment report is scheduled for completion in late 1998.

    Focus of this paper is on ozone trends derived from re-evaluated total ozone data obtained during 1979-1996 at 16 U.S. National Oceanic and Atmospheric Administration/Climate Monitoring and Diagnostics Laboratory (NOAA/CMDL) and cooperative Dobson spectrophotometer observatories. Seven of the stations (Table 1) are located on the U.S. mainland: Caribou, Maine; Bismarck, North Dakota; Boulder, Colorado; Wallops Island, Virginia; Fresno, California; Nashville, Tennessee; and Tallahassee, Florida. Two are lower-latitude CMDL stations: Mauna Loa Observatory (MLO), Hawaii, and Samoa Observatory (SMO), Tutuila Island, South Pacific. Four are foreign cooperative stations: Huancayo Observatory, Peru, operated by the Instituto Geophysico Del Peru; Haute Provence, France, operated by the University of Reims, Spectromtrie Molculaire et Atmosphrique, Champagne-Ardenne; Lauder, New Zealand, operated by the National Institute of Water and Atmospheres; and Perth, Australia, operated by the Australian Bureau of Meteorology. And three are CMDL polar region stations: Poker Flat, Alaska; Point Barrow, Alaska; and Amundsen-Scott (South Pole), Antarctica. In February 1993 the Poker Flat station was relocated to nearby Fairbanks. Alaska (64.9N, 147.9W), and in March 1995, the Fresno station was relocated to nearby Hartford, California (36.3N, 119.6W).

The Data and Trend Analysis

    NOAA total ozone data used herein for trend derivations are unique in that they have been obtained completely independently of other ozone measurement systems, and because calibrations of all Dobson instruments with which observations were made are traceable to one standard instrument, namely, World Primary Standard Dobson Instrument 83. Instrument 83 was established as a standard spectrophotometer in 1962 [Komhyr et al., 1989] for use in periodic calibrations of field instruments of the U.S. Dobson instrument station network, at which time, also, standard operating procedures defined by the International Ozone Commission were implemented, and sanctioned in 1980 by the WMO for use throughout the world. The long-term ozone measurement precision of, instrument 83 has been maintained since 1962 at 1%. This has been accomplished through Langley type calibration observations made periodically with the instrument at MLO and by means of standard lamps.

    Throughout the years provisional total ozone data from the station network have been routinely archived at the WMO World Ozone and Ultraviolet Radiation Data Centre (WOUDC) in Downsview, Ontario, Canada. To optimize data quality, a program was undertaken in 1992 to re-evaluate the data from all stations observation by observation. As part of this effort, a section was prepared for a WMO Dobson Data Re-evaluation Handbook [WMO, 1992] that provides detailed instructions for re-evaluating total ozone data having well-documented calibration histories such as those available for the NOAA record. Work in re-evaluating the more than 400 station-years of data is complete through mid 1995 for most stations in operation dining 1979 (Table 1), but through mid 1994 for shorter-record stations where observations began in the 1980s. (Tallahassee data, where observations were made sporadically throughout the years, have been only tentatively re-evaluated through 1989.) Several field instrument re-calibrations during 1996 and 1997, as well as scrutiny of recent-year measurement and calibration data from all stations, revealed minor errors only in Boulder and Perth data that were corrected. Maximum errors in the re-evaluated data total ozone monthly means are estimated not to exceed 1.5% taking the calibration scale of Primary Standard Dobson instrument 83 as standard. The re-evaluated data have been archived at the WOUDC.

    A regression-time series model [Reinsel et al., 1994] for monthly average total ozone time series data was used as the basis for seasonal trend analyses, accounting for solar cycle and QBO effects. Annual and seasonal (Dec.-Feb.; Mar.-May; June-Aug.; and Sept.-Nov.) ozone trends in percent ozone change per decade, derived from the linear regression model for the 16 stations under consideration, are shown in Table 1. Data records for the stations, in the form of deseasonalized total ozone monthly means, are plotted in Figures 1 and 2. Although several of the station data records date back to the early 1960s, tile focus on the trend determinations herein has been data obtained since 1978, a time period of significant downward trends in ozone at middle and high latitudes.

Results and Discussion

Table 1.    Annual and Seasonal Ozone Trends (in % per Decade) Derived from Re-Evaluated Dobson Spectrophotometer Total Ozone Data Obtained During 1979-1996 at 16 NOAA and Cooperative Stations, Allowing for Solar and QBO Equatorial Wind Effects
Station Latitude Longitude Period of Record Annual Dec.-Feb. March-May June-Aug. Sept.-Nov.
Trend s.e. Trend s.e. Trend s.e. Trend s.e. Trend s.e.
Wallops Island
Mauna Loa
Poker Flat*
Point Barrow*
Haute Provence
71.3N 43.9N

Average of first five stations (1979-1996)
Average of first five stations (1979-1995)
Average of first five stations (1979-1994)
Average of first five stations (1979-1993)
Estimated trend uncertainties are standard errors (s.e.) where 2 (s.e.) ~95% confidence interval.
*For Poker Flat and Barrow, the annual trend estimates are based on March through October data only, while "Sept.-Nov." trend estimates are based on September-October data only.
**For Amundsen-Scott, the annual trend is based on daylight (October 15 through February) data only. The "Mar.-May" trend is for Mayand June data only, the "June-Aug." trend is for July and August data only, and the "Sept.-Nov." trend is for October15-31 and November data only.

    At the U.S. mainland stations having essentially uninterrupted data records, namely, Caribou, Bismarck, Boulder, Wallops Island, and Nashville, ozone trends during 1979-1996 (Table 1) averaged -3.4, -4.9, -2.6, and -1.9%/decade, respectively, for winter, spring, summer, and autumn months, with the annual average trend being -3.3%/decade. Table 1 lists, also, average trend data for these five stations derived for 1979-1993, 1979-1994, and 1979-1995. Because record low ozone occurred over the U.S. during 1992-1993 [Komhyr et al., 1994] and most other parts of the world [Gleason et al., 1993], the linear trends derived from the 1979-1993 data, particularly for winter and spring months, are in part overestimates of the true long-term trends. With addition to the station records of data for 1994, 1995, and 1996, the results of the linear trend analyses procedure became less sensitive to the large, non-linear 1992-1993 ozone perturbation, yielding improved long-term 1979-1996 trend estimates for winter and spring months and on an annual basis that were less negative, respectively, by 1.5, 0.8, and 0.6%/decade than the trends deduced for 1979-1993.

    For comparison with the NOAA 1979-1994 five-station trend data, trends were derived, also, from December 1978-November 1994 Nimbus 7/Meteor 3 TOMS Version 7 satellite overpass total ozone data for the sites. They were determined to be -4.14, -5.83, -2.09, and -2.12%/decade for winter, spring, summer, and fall seasons, respectively, and -3.69%/decade on an annual basis. These results agree with NOAA trend data to within about 0.5%/decade (Table 1), with the NOAA trends being slightly more negative for winter but slightly less negative for spring months.
    Because of the shortness (1984-1996) of the Fresno ozone data record and the episodic nature of the Tallahassee record (Figure 1), ozone trends from these two stations should not be compared directly with trends at the five other U.S. mainland stations. While the annual ozone trend at Fresno of -3.7%/decade approximates the five-station average annual trend for 1979-1996 of -3.3%/decade, a difference is evident in the seasonal dependence of the trends. At Tallahassee, ozone trends derived from the episodic data approximate those at Fresno-a result that may be in part fortuitous, but may also reflect a real difference in the regional pattern of the trends.

Figure 1.  Deseasonalized total ozone anomaly data for 15 NOAA/CMDL and cooperative stations formed by subtracting 1979-1996 monthly normals from total ozone monthly means for the indicated periods of record. Observations at Point Barrow and Poker Flat were made only during the months of February to October of each year.

    Annual downward trends in ozone during 1979-1996 at the lower latitude Mauna Loa and Samoa Observatories were smaller, viz., -0.4 0.7 and -1.3 0.6%/decade, respectively, being not significant at Mauna Loa and barely significant at Samoa at the 95% confidence interval level. The annual trend derived from the shorter 1979-1992 Huancayo Observatory record was more negative (-1.9 0.4%/decade). This result is biased, however, by the pronounced ozone low that occurred at Huancayo during the 1992 austral winter (Figure 1). Excluding the end-of-record 1992 data, the 1979-1991 annual ozone trend for Huancayo was determined to be -0.9 0.4%/decade.

    NOAA cooperative total ozone observations began at Perth and Lauder in the Southern Hemisphere, respectively, in 1984 and 1987. Through 1996, both of these shorter-record stations show maximum downward trends in ozone of about -3.0 1.0%/decade occurring during austral summer months. On an annual basis, ozone trends at both stations were about 1.0%/decade, though not statistically significant and of opposite sign. Positive, not statistically significant trends occurred also during austral winter at Perth and during austral spring at Lauder. The large, positive, statistically significant wintertime trend of 5.0 2.3%/decade deduced for Lauder is, most likely, a reflection of the shortness (10 years) of the Lauder record and the anomalously high ozone values (not shown) present in the Lauder region during July months of 1994-1996, and August months of 1994 and 1996.

    Observations at the high-latitude North American stations of Poker Flat and Point Barrow were not made during winter months, nor at Point Barrow during 1983-1986. However, sufficient observations were made at both stations during spring and summer months to yield statistically significant ozone trends averaging -6.6%/decade during spring months and -3.5%/decade during summer months at the two stations. Autumn season data, represented primarily by observations made during September, yield a non-statistically significant mean trend for the two stations of -1.2%/decade.

    Ozone trends derived from the 1984-1996 Haute Provence (44N latitude) data exhibit the same seasonal variations present in the trend data of the mainland U.S. midlatitude stations. Statistically significant trends at Haute Provence for winter and spring months of -5.3 and 8.0, respectively, are, however, more negative by about 2.4%/decade than the average trends for corresponding seasons for Caribou, Bismarck, and Boulder, located at a mean latitude of 45N. This is most likely a consequence of the well known longitudinal variation in ozone trends in the Northern Hemisphere [see, e.g., Stolarski et al., 1992].

    As indicated earlier, heterogeneous chemistry involving the activation of chlorine molecules on the surfaces of polar stratospheric clouds has been the primary cause of ozone depletion in Antarctica during late winter and spring months. More recently, sulfate particles from the Pinatubo volcanic eruption have contributed to the ozone loss [Hofmann and Oltmans, 1993] South Pole 1979-1996 daylight (October 15- February) total ozone anomaly data plotted in Figure 2 exhibit the largest downward trend in ozone during 1979-1995, followed by a considerably reduced rate of ozone decrease during 1986-1995, with lowest ozone values occurring in 1996. The linear trend approximation of the ozone decrease rate for the 1979-1996 daylight time period as-a-whole is -12%/decade (Table 1). This is roughly one-half the downward trend in ozone (-20%/decade) deduced for spring (October and November) months when, on a seasonal basis, total ozone amounts at South Pole have been lowest. During austral summer months (December-February), the 1979-1996 South Pole ozone trend has been -7%/decade, a value twice that deduced on average from summertime (June-August) total ozone data obtained at the high latitude Northern Hemisphere stations of Poker Flat and Point Barrow.

Figure 2. South Pole station monthly total ozone anomalies derived from austral daylight observations made during October to February months.

    Moonlight, wintertime total ozone observations were made at South Pole throughout the years primarily during the months of April-August. Although observational frequency each month was limited to about one week's measurements when the moon was more than one-half full, trends deduced from the data appear to be meaningful since the natural variability of ozone confined within the Antarctica polar vortex during winter months is low. Note (Table 1) that the 1979-1996 negative ozone trends at South Pole for late autumn and winter months strengthened with seasonal progression, being -6%/decade for May-June months, but -10%/decade for the months of July and August.

    To compare ozone changes described above that occurred over the U.S. mainland during 1979-1996 with earlier data, annual and seasonal ozone trends were computed (not shown) from re-evaluated NOAA/CMDL data from Caribou, Bismarck, and Nashville dating back to 1963. The average annual 1963-1978 ozone trend for the three stations was determined to be -0.7%/decade. From the pre-1979 and post-1978 trend values, as well as from trends for the two time intervals determined from analyses incorporating the 1963-1996 data sets as-a-whole, a best estimate was determined for the annual ozone decrease present at the U.S. mainland midlatitude stations through 1996 relative to mid-1960s total ozone amounts, namely, -6.7%. Similar analyses, incorporating re-evaluated South Pole data obtained since 1961, yielded an ozone change at South Pole through 1996 of approximately -25% (daylight observations). Lowest total ozone amounts during 1979-1996 occurred at South Pole in October. Figure 3, which plots October 15-31 mean data, shows that, on average, October total ozone during the 1990s was lower than 1960s total ozone by about 50%. Lowest daily October ozone values are also plotted in Figure 3. A record daily ozone low of 99 DU was observed at South Pole on October 18, 1993.

Figure 3. South Pole station October 15-31 total ozone "monthly" means for 1964-1996 (circles). Crosses indicate October ozone minimum values. The dots plot wintertime (March-August) total ozone amounts.

Acknowledgments.  Prior to 1989, ozone observations described herein, were conducted by the NOAA Air Resources Laboratory and predecessor organizations. Appreciation is expressed to observers, too numerous to mention, at U.S. National Weather Service sites and at the other stations. Measurements at Huancayo, Haute Provence, Lauder, and Perth were made under supervision of J.A. Bravo, A. Barbe, W.A. Matthews, and J. Easson, respectively. Observing programs at Amundsen-Scott, Poker Flat/Fairbanks, Tallahassee, and Wallops Island were conducted in cooperation and with the Support of the U.S. National Science Foundation, the University of Alaska, Florida State University, and the National Aeronautics and Space Administration, respectively. G.L. Koenig of NOAA/CMDL assisted with data processing. Financial support for ozone data re-evaluation was received from CMDL, NESDIS, and the U.S. Department of Energy.


Bojkov, R.D., L. Bishop, and V.E. Fioletov, Total ozone trends from quality-controlled ground-based data (1964-1994), J. Geophys. Res. 100,25,867-25,876, 1995.

Gleason, J.F., P.K. Bhartia, J.R. Herman, R. McPeters, P.C. Newman, R.S. Stolarski, L. Flynn, G. Labow, D. Larko, C. Seftor, C. Wellemeyer, W.D. Komhyr, A.J. Miller, and W. Planet, Record low global ozone in 1992, Science, 260, 523-526, 1993.

Harris, N.R.P., G. Ancellet, L. Bishop, D.J. Hofmann, J.B. Kerr, R.D. McPeters, M. Prendez, W.J. Randel, J. Staehelin, B. H. Subbaraya, A. Voltz-Thomas, J. Zawodny, and C.S. Zerefos, Trends in stratospheric and free tropospheric ozone, J. Geophys. Res., 102, 1571-1590, 1997.

Hofmann D.J., and S.J. 0ltmans, Anomalous Antarctic ozone during 1992: Evidence of Pinatubo volcanic aerosol effects, J. Geophys. Res., 98, 18,555-18,561, 1993.

Komhyr, W.D., R.D. Grass, and R.K. Leonard, Dobson spectrophotometer 83: A standard for total ozone measurements, 1962-1997, J. Geophys. Res., 94, 9847-9861, 1989.

Komhyr, W.D., R.D. Grass, R.D. Evans, R.K. Leonard, D.M. Quincy, D.J. Hofmann, and G.L. Koenig, Unprecedented 1993 ozone decrease over the United States from Dobson spectrophotometer observations, Geophys. Res. Lett., 21, 201-205, 1994.

McPeters, R.D., S.M. Hollandsworth, L.E. Flynn, J.R. Herman, and C.J. Setter, Long-term ozone trends derived from 16-yr combined Nimbus 7/Meteor 3 TOMS Version 7 record, Geophys. Res. Lett., 23, 3699-3702, 1996.

Reinsel, G.C., G.C. Tiao, A.J. Miller, D.J. Wuebbles, P.C. Connell, C.L. Mateer, and J.J. DeLuisi, Statistical analysis of total ozone and stratospheric Umkehr data for trends and solar cycle relationship, J. Geophys. Res., 92, 2201-2209, 1987.

Reinsel, G.C., G.C. Tiao, D.J. Wuebbles, J.B. Kerr, A.J. Miller, R.M. Nagatani, L. Bishop, and L.H. Ying, Seasonal trend analysis of published ground-based and TOMS total ozone data through 1991, J. Geophys. Res., 99, 5449-5464, 1994.

Solomon, S., Progress towards a quantitative understanding of Antarctic ozone depletion, Nature, 347, 347-354, 1990.

Stolarski, R.S., P. Bloomfield, R.D. McPeters. and J.R. Herman, Total ozone trends deduced from Nimbus 7 TOMS data, Geophys. Res. Lett.,18, 1015-1018, 1991.

WMO, Handbook for Dobson ozone data re-evaluation, WMO Global Ozone Research and Monitoring Project Report No. 29, Geneva, 1992.

    1EN-SCI Corporation, Boulder, Colorado
W.D. Komhyr, EN-SCI Corporation, P.O. Box 3234, Boulder, CO 80107 (e-mail: 76353.551@compuserve.com)

    2University of Wisconsin, Madison

    3Cooperative Institute for Research in Environmental Sciences. Boulder, Colorado

    4Evolving Systems, Incorporated, Denver, Colorado

    5Formerly at NOAA Climate Monitoring and Diagnostics Laboratory. Boulder, Colorado 80303

Back to top

Paper number 97GL03313.