11.1     Introduction

          Ozone provides by far the most dominant absorption in the near UV spectrum, which is as it should be, of course, if we are to measure ozone with any accuracy in this region. At the same time, there are other atmospheric gases which absorb in the near UV, and hence which can interfere with the measurement, as it is defined by equation 1.4. The most important of these are sulphur dioxide (SO2), and nitrogen dioxide (NO2) (Komhyr and Evans, 1980). There also exists, in principle, the possibility of interference from UV emissions by atmospheric gases and particles.

          The effect of an interfering absorbing gas depends on its spectral absorption cross sections, and on the gas column amount and its spatial and temporal variation. For most atmospheric gases these quantities are not sufficiently well known for our purposes. In general, the spectral cross sections will be less accurately known than those for ozone, i.e., will have uncertainties of >>10%, and probably even larger uncertainties will be present for the relative absorption coefficients of paired bands. A coefficient in atm-cm-1 is multiplied by 3.722 x 10-20 to give the equivalent cross section in cm2, and by 1/0.4462 to give the equivalent in m2mol-1.

          The spatial distributions of gases range from being uniform, as occurs for globally diffuse sources or unreactive species, to being very variable, as occurs for point sources and reactive species. An interfering gas may be of natural origin or of anthropogenic origin. The effect at any site due to a nearby point source, such as a factory, city or volcano, will depend on the direction and the horizontal turbulent dispersion characteristics of the prevailing wind, and on any chemical loss (or gain) processes within the plume. The vertical extent of dispersion is of no direct interest since our concern is only with column amounts of the gas, i.e., height integrated amounts. However, column amounts can be deduced from concentration measurements if sufficient information on the height profile of the gas is available. For example, in urban situations, concentration data on various species are often available, and the scale height of the gas, or the height of the inversion layer which generally accompanies urban air pollution episodes, can be found from meteorological observations. The column amount x2 of a well mixed layer of gas is related to the mean concentration C in ppmv and the inversion height z by

                    (1 - exp-(z/H))
      x2 (atm cm) = ---------------- .  C(ppmv)                                           (11.1)

where H is the atmosphere's scale height, which equals 8.49 km for a layer at 17°C.

          Extremely high column amounts of an interfering gas can occur close to its source under conditions of very small horizontal dispersion, but of course these will have only very limited spatial extent, by definition. When a large number of point sources are spread over an extended region they may constitute an "area source", within which the wind direction and dispersion have little effect on the column amount experienced.

          The primary source of anthropogenic SO2 is the burning of sulphur-bearing fuels at large oil refineries, smelters, coal and oil fired electric power plants, and at other smaller industrial plants. Domestic coal and oil use may be a problem in some areas. The production of anthropogenic NO2, and the consequent local photochemical production of ozone is largely due to motor vehicle exhausts and is thus largely confined to urban areas. The production of other industrial waste gases, such as H2O2, is also confined largely to industrial and urban areas. The significance of large scale burnings of forest and of crop detritus, which occurs throughout the tropical and temperate land areas, is as yet unknown.

          The ozone error, ΔX, due to an interfering absorption is:

      ΔX (atm cm) =  ---- x2 (atm cm)                                                     (11.2)

(Komhyr and Evans, 1980), where α1 and α2 are the respective absorption coefficients of the ozone and the interfering gas for the particular band or band combination of interest. Thus the errors will not show any airmass dependence, except if the amount of interfering gas has a diurnal variation, which it well might.

          If absorption interference due to point sources is a persistent problem, then it will almost always be possible to relocate the Dobson instrument to a site distant from, and predominantly upwind of, the offending source. A shift of even 10 km may have a significant effect, though to avoid the plum of a city a shift of 100 km or more may be needed (Komhyr and Evans, 1980). Each Dobson site needs to be considered individually for its pollution potential. When the problem is an area source, it may be necessary to correct for the interference by means of supplementary chemical and meteorological measurements.

          Some general considerations show that the effects of atmospheric radiation emissions in the UV will be very small. Fluorescence derives its energy from the absorption of more energetic photons, i.e., from light of shorter wavelength. However, light of wavelengths less than 300 nm will not penetrate the ozone layer to reach the bulk of the atmosphere where presumably the bulk of any fluorescing species will reside. Any fluorescence from above the ozone layer necessarily will be weak and will be absorbed in passing through the ozone layer to an extent similar to that for extraterrestrial sunlight. Also, the fluorescent energy is emitted into a full 4π steradian field, only a small part of which is received by the Dobson field of view, and fluorescence (and luminescence) is not an efficient energy conversion process. Luminescence (airglow) is generally very weak, as is evidenced by the degree of blackness of the rural night sky.

          Although it is possibly out of context here, it is worth mentioning the so-called Ring effect, which refers to the reduction in the depth of the deep and very narrow Fraunhofer lines in the near UV and visible skylight relative to the same lines in direct sunlight. The effect amounts to a filling in of a few percent (Brinkman, 1968). A number of explanations have been proposed, including airglow, aerosol fluorescence and Raman scattering, but none is as yet generally accepted (Kattawar et al., 1981). The important consideration for our problem is that any effect on the flux for the relatively broad 1 nm Dobson bands will be very small, almost certainly less than 0.5%.

11.2     Interfering absorption

          Dobson (1963) briefly discussed the question of interfering absorption and concluded that only SO2 would be a problem, and that even then, its effect was negligible for his instrument at Oxford. Komhyr and Evans (1980) studied the effects of a range of anthropogenic pollutants known to have UV absorption bands, namely SO2, NO2, N2O5, H2O2, HNO3, acetaldehyde, acetone, acrolein, and locally produced ozone. For each pollutant they estimated, firstly, the absorption coefficients at the Dobson bands, and secondly, the likely column amounts from surface concentration data and mean mixing height or inversion height estimates. Only SO2 and NO2 showed any significant effects. In clean air their column amounts are of the order of 0.0001 atm cm, but in the most extreme cases their concentrations can approach 1 ppm, which if present throughout a 0.5 km layer, would result in column amounts of 0.046 atm cm. For this most extremely polluted air, Komhyr and Evans (1980) calculate errors in the AD ozone estimation of 25% and 5% respectively for SO2 and NO2, and corresponding errors in the CD ozone estimation of 45% and 11%.

          The above worst-case errors are very large, but in practice their occurrence will be very limited. Most Dobson instrument sites are not subject to high pollution, and the time averages of the errors, say over a month, will be considerably less than the peak values. For example, Komhyr and Evans (1980) estimate for a site 16 km from Denver, in an area near two oil refineries and an electrical generating plant, that AD errors due to SO2 would have a maximum of 15% (three hour average), would exceed 3% on perhaps four days per year, and would amount to only about 0.6% when averaged over the year. Similarly, they estimate for Pasadena, 16 km from Los Angeles, that AD errors due to NO2 would have a maximum of 3% (one hour average), and would amount to 0.8% for the annual average of the daily maximum, and to 0.5% for the annual average of all the data. Thus, even for such chronically polluted sites, the errors in long term AD ozone data due to urban SO2 and NO2 will be only about 1%.

          Komhyr and Evans (1980) show that the combined error contribution of N2O5, H2O2, HNO3, acetaldehyde, acetone and acrolein for highly polluted air is less than 0.3% and is therefore negligible. Photochemically produced ozone in polluted air is part of the ozone column, but it is not part of what one might consider to be "background" or "naturally produced" ozone, and it certainly could interfere with efforts to understand long term variations of stratospheric ozone amounts. Komhyr and Evans (1980) estimated that in very extreme pollution, this locally produced ozone could add about 0.025 atm cm, or about 8%, to the ozone column amount, but that the time-averaged errors for even chronically polluted sites are probably less than 1%. Most sites will never be affected.

          Direct measurements of column amounts of SO2 have been made by Evans et al. (1981) and Kerr et al. (1981) using the Brewer spectrophotometer. With measurements from its five bands, it is possible to simultaneously solve for the ozone column amount, the SO2 column amount, and the local spectral gradient of aerosol attenuation, provided, of course, that there is no other interfering absorption present. Measurements for the period September 1979 to June 1980 at Toronto showed daily average values of SO2 of up to 0.010 atm cm and a long term average of about 0.003 atm cm. Using Komhyr and Evans (1980) absorption coefficients, these imply AD ozone errors of about 5% and 1.5% respectively. These are very approximately twice those estimated for anthropogenic SO2 at Denver by Komhyr and Evans (1980).

          Evans et al. (1981) had the good fortune to experience on 20 and 21 May 1980, the overhead passage of the Mt. St. Helens eruption plume, and to thereby measure its elevated column amounts of SO2. The maximum amount measured was in excess of 0.050 atm cm, which would have resulted in an error in the AD ozone estimate in excess of 0.075 atm cm. The SO2 column amount for this event could also be estimated from the Dobson instrument measurements and these SO2 estimates agreed well with those from Brewer instrument measurements. Also, the Dobson ozone estimates, when corrected for the SO2 interference, agreed well with the (implicitly corrected) Brewer ozone estimates. Overall, the work of Evans et al. (1981) and Kerr et al., (1981) provides a picture of SO2 interference which is consistent, at least within the uncertainties inherent in the measurement techniques.

          The Mt. St. Helens measurements have led some investigators to the view that small but important levels of interference will be present due to stratospheric SO2 and its periodic replenishment by volcanic eruptions, since although the volcanic emissions of SO2 are small relative to anthropogenic SO2 emissions (e.g. Sigurdsson (1982) quotes yearly values of 2.8 x 105 tons and 1.3 x 108 tons respectively), the residence times of SO2 in the stratosphere are greater than those in the troposphere. Thus there may be globally distributed stratospheric SO2 levels of the order of 0.005 atm cm with residence times of the order of many months. This implies slowly varying errors in the Dobson AD ozone estimates of the order of 1%. This view is supported by an analysis of Brewer instrument measurements made before and after the recent January 1982 stratospheric "Mystery Cloud" (W.F.J. Evans, personal communication) and by the intense and widespread interference observed in the satellite SBUV/TOMS ozone estimates in April 1982 (A.J. Krueger, personal communication) in the vicinity of the plume of the erupting Mt. El Chichon in Mexico.

          The significance of a volcanic event depends on the amount of sulphur injected into the stratosphere and hence upon the sulphur content of the erupting rock and the degree to which the eruption penetrates the tropopause. These factors will vary considerably but probably they could be estimated for the major eruptions of this century. A large part of the initial sulphur input could be in the form of sulphur compounds other than SO2. The effect of some eruptions may be considerably larger than that of Mt. St. Helens. For example, it seems that the El Chichon eruption was much larger, and Sigurdsson (1982) reports that the massive Laki eruption of 1783 produced 5 x 107 tons of SO2, which was approximately 300 times the Mt. St. Helens SO2 production.

          The effects, if any, of other atmospheric gases on the Dobson measurements have not been studied explicitly it seems. The work of Thompson et al. (1963) and the review of Hudson (1974) imply that CO2, CO, O2, H2O, N2O, NH3, and NO have no absorption bands in the 300 to 350 nm region. The chlorofluorocarbons CF2Cl2, CFCl3, and CCl4, and presumably also their bromocarbon counterparts, do not absorb at ultraviolet wavelengths greater than 290 nm (Molina, 1980). There are a number of trace constituents, such as BrO, ClONO2, and HOCl which do absorb in this wavelength range (Molina, 1980), but in these cases it is quite probable that their column amounts are too small to cause any significant interference. The possibility of interfering absorption from molecular complexes, ranging from dimers to aerosol particles, has yet to be properly considered. The dimer (NO2)2 is known to absorb in the UV (Molina, 1980), as is the collision pair (O2 )2 (Perner and Platt, 1980). The UV absorption of aerosols is much smaller than their infrared absorption (Toon and Pollack, 1976), but it should not be neglected as a potential source of interference.

          Anomalous spectral features have been found in various UV spectral intensity measurements and have given rise to speculation as to possible interfering species. Kerr (1973) discussed nitrogen hydride, NH, as a possible candidate for explaining anomalies at Dobson and Brewer bands, though at that time NH concentrations were unknown and since then no further work has been reported. Krueger (1969) considered the effect of excited oxygen molecules in an attempt to explain discrepancies in his high altitude rocket optical ozone measurements. DeLuisi (1980) and McPeters and Bass (1982) note the possibility of absorption interference in seeking to explain anomalies in their UV spectra of transmitted radiation and satellite backscattered radiation respectively. It must be pointed out, however, that all these measurements will be limited by the uncertainties in the ozone absorption spectra used in analysing the intensity spectra.

          The only firm direct evidence of significant absorption interference is that for SO2, but at the same time there is a multitude of atmospheric gases whose absorption spectra and column amounts are at best poorly known. Even with SO2, the absorption coefficients used for Dobson bandpairs are uncertain to 30% (Evans et al., 1981) and are probably temperature dependent, and our knowledge of its column amounts are still very limited. Clearly, there is a great need for more comprehensive studies of the problem. Particular emphasis should be given to the analysis of high quality transmission or backscattered spectra, of the sort done by DeLuisi (1980) and McPeters and Bass (1982), and using the new Bass and Paur (1981) absorption coefficients. Transmission spectra for long horizontal paths would also be very useful. Emphasis should also be given to the study of the effects of volcanic eruptions, and possibly also of large scale fires of forests and crop material.

11.3      Interfering emission

          Very little has appeared in the literature on the subject of atmospheric emissions in the near UV region, which is probably indicative of its overall insignificance. As was discussed in the Introduction, 11.1, there are fundamental reasons why this is likely to be so.

          Kulkarni (1968) discusses the question of UV airglow interfering with Dobson B bandpair measurements made on the moon and concluded that the Herzberg band of molecular oxygen and the hydroxyl radical, OH, are possible sources. In their search for explanations for anomalous backscattered UV spectral features, McPeters and Bass (1982) suggested NO emission as the cause for a band at 272 nm, but they considered that the only possible source for emission in the 300 to 310 nm region was oxygen, and even then the intensities required would be unreasonably large. Aerosols might possibly be a source of fluorescent emission. The backscattering of night lighting by aerosols and air molecules, though not an emission, might be a small source of interference to moonlight measurements, especially when the ground is snow covered. In total, the above evidence is weak and does not point to any significant interference.

11.4     Summary

(i)      Absorption by atmospheric constituents other than ozone can interfere with the accuracy of Dobson ozone estimations. Of the multitude of anthropogenic, volcanic and background gases or particles, only a few are thought at present to interfere in any significant way, but our knowledge of the relevant column amounts and absorption cross sections is far from complete.

(ii)     Column amounts of anthropogenic SO2 in extremely polluted air may cause overestimates approaching 25% in AD ozone estimates. However, even at chronically polluted sites, the errors would exceed 5% on only a few days per year and would average to 1% or less over a year.

(iii)    Column amounts of anthropogenic NO2 in extremely polluted air may cause overestimates approaching 5% in AD ozone estimates. However, even at chronically polluted sites, the errors would exceed 3% on only a few days per year and would average to 1% or less over a year.

(iv)     In very polluted air, the total error in AD ozone estimates due to maximum amounts of the anthropogenic known UV absorbers N2O2, H2O2, HNO3, acetaldehyde, acetone and acrolein is less than 0.5%. The presence of other interfering species cannot be ruled out. The effect of large scale anthropogenic burnings of forest and crop detritus is unknown.

(v)      Photochemically produced ozone in polluted urban air can interfere with the measurement of the "background" column ozone amount. On occasion it may rise to 8% of the column total, but even at chronically polluted sites the long term contribution will be less than 1%.

(vi)     Volcanic eruptions may produce sufficient SO2 to cause errors of 30% or more to AD ozone estimates made beneath the plume up to some hundreds of kilometres from the volcano. Also, sufficient SO2 may be injected into the stratosphere by volcanic eruptions to cause AD ozone errors of 1 to 2% over regional or global scales which decay slowly over periods of many months.

(vii)    The combination of theoretical and experimental evidence indicates that UV radiation emission within the atmosphere is a negligible source of error.

(viii)   Overall, the error due to interfering absorption for yearly average AD ozone estimates is likely to be less than 2% for chronically polluted sites and less than 0.5% elsewhere. Spatially averaged errors will tend to be even less. Errors for the CD estimation are approximately twice those of the AD combination.

Return to Table of Contents

Forward to Airmass Accumulation