Search

A new UV–visible spectrometer system that measures the absorption of light from stars and planets by constituents in the Earth’s atmosphere is described. Because it can be used to make measurements at night, the system has a significant advantage for measuring polar constituents in winter, when conditions that might give rise to ozone loss are initiated. Other advantages arise from the use of a cooled two-dimensional CCD array as the detector: an array detector avoids spectral noise resulting from scintillation of stars or from clouds passing overhead and allows for the possibility of measuring several constituents simultaneously; its second dimension permits auroral light from the atmosphere adjacent to the star to be measured simultaneously and subtracted from the stellar light, and a modern low-noise CCD allows us to use a telescope of modest diameter. The few previous measurements of constituents made by the use of stellar absorption did not have these advantages. The instrument was configured for simplicity and ease of use in field measurements and was deployed outside in winter in Northern Sweden in 1991. Examples of ozone measurements are shown.

Visible spectra of the zenith-sky were recorded at Faraday in Antartica (65°S) during 1990. Line-of-sight ozone amounts were deduced by fitting laboratory cross-sections to the spectra. At the lattitude of Faraday, visible spectra are more useful than the u.v. wavelengths used by the Dobson spectrophotometer, because they permit observations of ozone throughout the winter when the sun is too low in elevation for useful measurements with the Dobson. In order to find the accuracy of the visible-light system for the determination of medium- and long-term trends in ozone, Langley plots of the ozone amounts were examined to determine the stability of the offset to the zero of the ozone measurement. The plots revealed changes which correlated with shifts in wavelength of the spectrometer. This allowed the use of wavelength shift as a surrogate for offset, fortunate because the use of Langley plots to determine the offset on an individual day from zenith-sky spectra at high latitudes is unreliable. The standard deviation of the ratios of total ozone measured on the same day at Faraday by the Dobson spectrophotometer and the visible spectrometer was 10% using offsets calculated from wavelength shifts. This is 4% less than the standard deviation using a straight-line fit to the offsets. Although common for Dobson measurements of ozone in the u.v., such Langley-plot analyses are, so far, unusual for visible spectrometers, but are essential for future trend determination, even by systems that are inherently more stable.

Measurements of stratospheric composition have been made with a novel star-pointing spectrometer. The instrument consists of a telescope that focuses light from stars, planets, or the moon onto a spectrometer and two dimensional CCD array detector. Atmospheric absorptions can be measured, from which atmospheric columns of several gases can be determined. The instrument was deployed in Abisko, 69 deg N, during the European Arctic Stratospheric Ozone Experiment (EASOE). The instrument has the potential for measuring O3, OClO, NO2, and NO3. In this paper, a method for the retrieval of vertical columns is described, and some examples of ozone measurements given.

The measurement of N2O5 by Evans [Appl. Opt. 25, 1866–1868 (1986)] contains several serious errors. A reanalysis of the published infrared spectra gives the more realistic error bar of ±45%.

Radiative transfer within an interferometric spectrometer is discussed in relation to measurements of infrared transmittance spectra of gases. It is shown that for certain types of interferometric spectrometers, radiation from sources other than the spectrometer source must be considered, including the gas sample itself, if significant errors in measurement of line strengths and line widths are to be avoided. The validity of the radiative transfer theory is demonstrated by applying it to spectra of an evacuated gas cell, recorded over a range of temperatures, and thereby deriving a temperature independent emissivity for the cell windows. The error introduced into measurements of spectral line strengths and line widths is discussed with reference to experimental values of these parameters obtained for lines in the v2 band of H2O using a Bomem DA 3.002 spectrometer. Ways in which the effects of radiation from the gas cell and its contents can be minimised or eliminated are discussed, including an experimental procedure involving measurements with two different temperatures of the spectrometer source.

In a new program which performs a Backus-Gilbert analysis of the output kernels from the input kernels of any hypothetical remote sensor, Conrath's method for calculating trade-off curves has been shortened and the problem of normalization of truncated input kernels has been addressed. When evaluating the performance of remote sensors at the design stage, it is useful to transform the units of the input kernels into those of the output, and a method for treating this transformation has been formalized. We also show that when output kernels are selected for maximum signal-to-noise ratio and the input kernels have a limited vertical range, the output kernels move towards the centre of the input range. This conclusion is not a property of the sensor being analysed.

Measurements of N2O5 in the stratosphere and of its spectrum at 1240 cm−1 in the laboratory are reviewed. Errors in previous publications about N2O5 are discussed briefly. The earliest stratospheric measurement is significantly revised in the light of new measurements of band strength. All the measured profiles are tabulated using a standard value for the band strength; they can be linearly scaled if future, more accurate, measurements suggest that a different value should be used