bandkl1 (58K)

mpilogo4 (4K)


Doas-History


History of UV/vis satellite remote sensing Spectroscopic analysis of electromagnetic radiation allows to retrieve information on the processes which controlled its release and transfer. For instance spectroscopy is thus a key method of remote sensing. Spectroscopy has been the main source of information in astronomy.

Since about 100 years it has also successfully been applied to study the atmosphere of our planet, of course long before that time the colour of the sky has inspired peoples imagination and environmental interests. In this section a short overview on the history of spectroscopic methods is given which are relevant for the UV/vis satellite remote sensing of the earth’s atmosphere. About 200 years ago Joseph von Fraunhofer (1787 - 1826) was for the first time able to produce spectroscopic gratings of so far unrivalled quality.

With these gratings he discovered numerous dark lines in the spectrum of solar light (Fig. 1). He reported: ‘I found very many strong and weak vertical lines, which are darker than the remaining part of the spectrum. Some of them are almost dark’.

Fraunhofer_Spektrum_von_1814_x (18K)

Fig. 1 Sketch of the solar spectrum coloured by Joseph v. Fraunhofer (around 1814). (©http://www.fmc.uni-karlsruhe.de/~timo/spektro1.html)

He was able to determine the exact wavelength of many ‘Fraunhofer lines’ as we call them today, and his nomenclature is still in use. Nevertheless, at that time he was not able to understand the physical basis and significance of his discovery, which nevertheless can be regarded as the first application of remote sensing to the atmosphere of a celestial body.

One important step forward in the understanding of the Fraunhofer lines was the discovery of the physical principles of absorption and emission in 1859 by the very fruitful collaboration of the physicist Gustav R. Kirchhoff (1824 - 1887) and the chemist Robert W. Bunsen (1811 - 1899) in Heidelberg. Using their ‘Spektralapparat’ (see Fig. 2) they were able to assign specific emission lines to different elements.

With this method they in particular discovered two new elements, Cäsium and Rubidium. They extracted very small amounts (7 grams) out of 44000 litres water from the mineral spring of Bad Nauheim (near Frankfurt/Main, Germany). In 1859 [Kirchhoff, 1859] they found that: ‘the vapour of table salt absorbs the same lines which it also emits. These lines are identical with solar Fraunhofer lines’.

Their discoveries constitute the foundation of two basic kinds of spectroscopy: spectroscopy of emission and absorption. In particular it was now possible to explain the solar Fraunhofer lines as absorption lines of elements in the solar atmosphere. Since then the spectroscopy of star light has become a very powerful tool in astronomy. For example, Helium was discovered in the solar atmosphere by the respective Fraunhofer lines before it was found on earth.

hdoas_fig2 (40K)

Fig. 2 Spektralapparat as build by Gustav R. Kirchhoff and Robert W. Bunsen, around 1823. (Chemical Achievers)

Spectroscopic measurements of the earth’s atmosphere Spectroscopic techniques were also applied to the terrestrial atmosphere, but did not only focus on atmospheric absorptions. In the higher atmosphere also emissions occur (e.g. polar light or nightglow from excited OH, O3, O2, and Na) which were studied with spectroscopic methods.

It was possible to identify exited atoms and molecules as sources of the observed radiation. The spectroscopy of atmospheric absorptions using sun (or moon) light, however, was complicated by the fact that most atmospheric absorption structures are usually by far weaker than the solar Fraunhofer lines.

Thus absorption spectroscopy of the earth’s atmosphere first concentrated only on strong absorptions like e.g. those of ozone, which are clearly identifiable in absorption spectra of solar light. (see Fig. 3).

hdoas_fig3 (14K)
Fig. 3 Spectra of direct sun light and light reflected by the earth observed by the satellite instrument GOME on board the European research satellite ERS-2 [ESA, 1995].

Besides many narrow solar Fraunhofer lines also a few atmospheric absorption features are visible: the strong Huggins O3 absorptions < 350 nm, and the weak Chappuis O3 absorption in the visible spectral range (see text). The atmospheric O2 absorption was already discovered by Joseph v. Fraunhofer but treated by mistake as a solar Fraunhofer line.

In 1879 Marie Alfred Cornu found that the short wavelength limit of the solar radiation on the earth’s surface must be caused by an absorber located in the earth’s atmosphere. One year later Sir Walther Noel Hartley described the strong UV O3 absorptions between 200 and 300 nm and it became obvious that the absorbing properties of O3 fulfilled the requirements of that postulated atmospheric absorber.

Further parts of the O3 absorption were discovered in the following years. In 1880 J. Chappuis discovered the much weaker visible absorptions (400 - 840 nm) in liquid ozone. In 1890 Sir William Huggins discovered the highly structured O3 absorption between 300 and 360 nm in spectra of Sirius. Because of their characteristic absorption feature these ‘Huggins Bands’ as they are now called are well suited for the identification and quantification of atmospheric O3 absorptions in solar spectra (see Fig. 4).

hdoas_fig4 (12K)

Fig. 4 O3 absorption cross section [Bass and Paur, 1985]. The highly structured O3 Huggins Bands are also displayed in the insert graph with an expanded wavelength scale. They are often applied for atmospheric O3 measurements.

In 1913 Albert Wigand concluded from balloon borne spectroscopic measurements over Berlin that the atmospheric ozone layer must be located above 9 km altitude. The exact altitude of the ozone layer was further constrained in the 1920s by Gordon M.B. Dobson and Paul Götz.

In 1925 Dobson developed a new very stable photospectrometer (double monochromator using quartz prisms) for the quantification of the atmospheric vertical column density (the vertically integrated concentration) of O3. Still today the ‘thickness’ of the atmospheric O3 layer is expressed as Dobson units (DU) which is the thickness (measured in 10-5 m) of the atmospheric O3 vertical column density under normal conditions (1 DU equals 2.68 ? 1016 molecules/cm²).

In 1926 Paul Götz investigated the variation of the intensity inside and outside ozone absorption bands during sunset. From the inversion of this intensity ratio (the so called ‘Umkehr-Effect’) he concluded that the maximum of the ozone layer must be located around 25 km. Finally, in 1934 Erich Regener was able to perform balloon flights into the stratosphere and he directly measured the UV absorption of the ozone layer.

The height profile of the atmospheric ozone layer was 1929 also confirmed by a (simple) photochemical theory involving only atmospheric oxygen introduced by Sidney Chapman at the first ozone conference at Paris. In addition, temperature measurements and energetic considerations confirmed the assumption of an absorbing layer above ca. 10 km. In 1902 Leon Teisserenc de Bort and Richard Aßmann independently discovered that the atmospheric temperature starts to increase above about 10 km (In 1908 Teisserenc de Bort called this layer between about 10 km and 50 km the stratosphere).

This temperature increase was in agreement with the assumption of an absorbing (O3) layer at that altitudes. After the discovery of the stratospheric O3 layer and its importance as UV-filter a nd thus for live on earth Dobson spectrometers became widely used; a global network of more than 100 instruments was established during the last 8 decades; most of them are still in use and provide long time series, which are the basis for the investigation of possible trends of the atmospheric O3 layer (see Fig. 5).

hdoas_fig5 (11K)

Fig. 5: Yearly averages of the O3 VCD at Arosa (Switzerland) observed by Dobson spectrometers (the black squares indicate high uncertainties). After 1970 a significant decrease (-20 DU/30 years) was determined. (© meteo-swiss, http://www.meteoschweiz.admin.ch/web/de/wetter/ozone_layer.html)

Dobson spectrometers use a very simple (but very stable) spectroscopic method: the direct or scattered solar intensity is measured in different narrow (about 1 nm) spectral intervals which are located either in or outside of O3 (Huggins) absorption bands (see Fig. 4 and 6).

From the ratio of the radiation intensity of such wavelength pairs ( and a geometric correction factor for the solar zenith angle, SZA) the vertical column density of O3 is determined. A possible influence on the intensity ratio due to the absorption of atmospheric aerosols can be accounted for by the combination of several wavelength pairs, for which the influence of O3 and aerosol is different.

In 1973 Alan Brewer used a similar method to measure the atmospheric NO2 column density from ground [Brewer et al., 1973]. He measured light in different narrow wavelength intervals around 450 nm, for which the NO2 absorption shows strong differential structures. Although the atmospheric NO2absorptions are by far smaller than those of O3, it was possible to clearly identify these absorptions with this new spectrometer. From this relatively simple measurement it was in particular possible to confirm the permanent existence of stratospheric NO2.

Two major steps forward in measuring weak atmospheric absorbers were introduced in 1975 by J. F. Noxon [Noxon, 1975]. First, he measured the intensity of the solar spectrum over a continuous interval in the spectral region where the atmospheric absorption appears (in this case around 440 nm for NO2).

Second, he removed the strong structures of the solar Fraunhofer lines by dividing a measured spectrum (during sunset or sunrise) by another spectrum measured during noon. Because of the much longer atmospheric light path at low sun this ratio contains strong absorption structures (the difference between twilight and noon) of the atmospheric absorber.



<
Dobson spectrometers use a very simple (but very stable) spectroscopic method: the direct or scattered solar intensity is measured in different narrow (about 1 nm) spectral intervals which are located either in or outside of O3 (Huggins) absorption bands (see Fig. 4 and 6).

hdoas_fig6 (11K)

Fig. 6 Various wavelength pairs used for O3 observations of Dobson Photospectrometers. Also displayed is the O3 absorption cross section in the Huggin bands. From the ratio of the measured intensities the atmospheric O3 column density is derived.

From the ratio of the radiation intensity of such wavelength pairs (and a geometric correction factor for the solar zenith angle, SZA) the vertical column density of O3 is determined. A possible influence on the intensity ratio due to the absorption of atmospheric aerosols can be accounted for by the combination of several wavelength pairs, for which the influence of O3 and aerosol is different.

In 1973 Alan Brewer used a similar method to measure the atmospheric NO2 column density from ground [Brewer et al., 1973]. He measured light in different narrow wavelength intervals around 450 nm, for which the NO2 absorption shows strong differential structures. Although the atmospheric NO2absorptions are by far smaller than those of O3, it was possible to clearly identify these absorptions with this new spectrometer. From this relatively simple measurement it was in particular possible to confirm the permanent existence of stratospheric NO2.

Two major steps forward in measuring weak atmospheric absorbers were introduced in 1975 by J. F. Noxon [Noxon, 1975]. First, he measured the intensity of the solar spectrum over a continuous interval in the spectral region where the atmospheric absorption appears (in this case around 440 nm for NO2).

Second, he removed the strong structures of the solar Fraunhofer lines by dividing a measured spectrum (during sunset or sunrise) by another spectrum measured during noon. Because of the much longer atmospheric light path at low sun this ratio contains strong absorption structures (the difference between twilight and noon) of the atmospheric absorber.


hdoas_fig7 (50K)

Fig. 7 Different steps of a DOAS analysis using the sun as light source.

First the Fraunhofer lines are (mainly) removed by dividing the measurement by a spectrum of direct sun light. The logarithm of this ratio is simultaneously fitted by the absorption cross sections of the trace gases which show differential absorption structures in the selected spectral range. Also included in the fitting procedure is a so called Ring-spectrum, which corrects the ‘filling-in’ of solar Fraunhofer lines by Raman-scattering [Grainger and Ring, 1962].

Usually, also the (logarithm of the) direct sun light spectrum is included in the fitting procedure. In this example the very weak atmospheric BrO absorption is determined from DOAS satellite observations (GOME) in the UV spectral range. The trace gas absorption spectra measured in the laboratory (red) are scaled to the respective absorptions in the GOME spectrum. The derived BrO absorption is by orders of magnitude smaller compared to the differential structures of the Fraunhofer lines (see numbers left of the y-axis).

In 1976 U. Platt and D. Perner introduced the DOAS (Differential Optical Absorption Spectroscopy) method [Perner et al., 1976; Platt et al., 1979; Perner and Platt, 1979]. The key concept of DOAS is the simultaneous fit of several trace gas absorption spectra to a measured atmospheric spectrum (see Fig. 7) [Platt, 1994].

This concept is the best way to avoid possible spectral interference between different absorption structures (e.g. like between SO2 and O3 for Dobson measurements). Today this concept i s basically still unchanged; however, several improvements were added (see also Platt et al. [1997]):

    • The correction of the ‘filling-in’ of the solar Fraunhofer lines by atmospheric Raman- scattering [Grainger and Ring, 1962; Solomon et al., 1987a; Wagner et al., 2001a].
      The consideration of the temperature dependence of atmospheric absorptions [e.g. Bass and Paur, 1985].
      The correction of a possible spectral shift and squeeze between the twilight and noon spectra [Solomon et al., 1987a; Stutz and Platt, 1996]
      The correction of spectral interference between atmospheric absorbers with the solar Fraunhofer lines because of the limited spectral resolution (in the order of 1 nm) of typical DOAS instruments, usually referred to as I0 effect [Johnston., 1996]
      The ’filling-in’ of atmospheric absorption lines by Raman scattering [Fish and Jones, 1995]
      The influence of strong atmospheric absorbers on the absorption structures of weak absorbers in atmospheric spectra (‘wavelength dependent air mass factor’) [Platt et al., 1997; Marquard et al., 2000]
      The correction of saturation effects for strong atmospheric absorbers which cannot spectrally resolved by DOAS instruments (e.g. H2O, O2) [Solomon et al., 1989, Wagner et al., 2000; Wagner et al., 2003a]
      The correction of instrumental problems (e.g. stray light in the spectrometer) during the DOAS fitting procedure [Aliwell et al., 2002]


  • hdoas_fig8 (11K)

    Fig. 8 Light paths through the atmosphere for DOAS observations of direct (top) or zenith scattered light (bottom).

    The Air mass factor (AMF) is defined as the ratio of the measured slant column density (SCD, the trace gas concentration along the light path) and the vertically integrated trace gas concentration (vertical column density, VCD). For direct light observations the AMF simply describes the geometric extension of the slant absorption path compared to the vertical path.

    Except for large solar zenith angles (SZA) the AMF can be approximated by the inverse of the cosine of the SZA. For scattered light observations, various light paths contribute to the measured spectrum and the AMF must be determined by modelling the radiative transport through the atmosphere.

    Usually the analysis of DOAS measurements (from ground, balloon or satellite) using an extraterrestrial light source (sun, moon, stars) includes two basic steps (see Fig. 8). First, from the DOAS fitting procedure the so called slant column density (SCD, the trace gas concentration integrated along the light path) is calculated. In a second step the radiative transport through the atmosphere is modelled. From such models the so called air mass factor (AMF) is determined, which is the ratio of the SCD and the vertically integrated trace gas concentration (VCD).

    Dividing the measured SCD by the AMF yields the atmospheric trace gas VCD. This t race gas VCD is the ‘classical’ quantity derived from DOAS measurements. If the (relative) atmospheric concentration profile shape is known even an absolute concentration profile can be calculated from the VCD. Limited profile information can also be extracted from DOAS observations themselves by applying three sophisticated methods:

    The investigation of the dependence of the measured trace gas absorption on the SZA (similar to the Umkehr-method for Ozone) [Solomon et al., 1987a; Preston et al., 1997] The investigation of the temperature dependence of the atmospheric trace gas absorptions, e.g. those of tropospheric and stratospheric NO2 [Richter, 1997; Richter et al., 1998a]. The recently developed so called Multi AXis (MAX) DOAS measurements are. They operate in several viewing directions [Hönninger and Platt, 2002], which have different sensitivities for different atmospheric altitudes.


    Atmospheric trace gas observations from Space

    hdoas_fig9 (23K)

    Fig. 9 The first image of the earth and its atmosphere was made in April 1960 from the American Television Infrared Observations Satellite TIROS 1. (© NASA, http://www.earth.nasa.gov/history/tiros/tiros1.html)

    First Satellite UV/vis observations simply showed pictures of the earth’s surface and atmosphere (see Fig. 9). Such satellite images are still used, for instance as input for numerical weather forecast. The first spectroscopic UV/vis observations started in 1970 on board of the US research satellite Nimbus 4 (see Fig. 10).

    These measurements (Backscatter Ultraviolet, BUV, later also called Solar BUV, SBUV) operated in nadir geometry; i.e. they measured the solar light reflected from the ground or scattered from the atmosphere. Like for the Dobson instruments (see Fig. 6) also the BUV/SBUV instruments measure the intensity in different narrow spectral intervals (Fig. 11). The intention of these BUV/SBUV observations was to determine information on the atmospheric O3 profile, since the penetration depth into the atmosphere strongly depends on wavelength.

    For example, the light at the shortest wavelengths has only ‘seen’ the highest parts of the O3 layer whereas the longest wavelengths have seen the total column. While in principle the BUV/SBUV measurements worked well, they suffered from instrumental instabilities. The big breakthrough in UV/vis satellite remote sensing of the atmosphere took place in 1979 with the launch the Total Ozone Mapping Spectrometer (TOMS) on Nimbus 7.

    TOMS is similar to the BUV/SBUV instrument but measures light at longer wavelengths (see Fig. 10, 11). Thus it is only sensitive to the total O3 column (instead of the O3 profile). However, compared to the BUV/SBUV instruments the TOMS instruments were much more stable. The TOMS instrument on board of Nimbus 7 yielded the so far longest global data set on O3 (1979 - 1992) [McPeters et al., 1996].

    This period in particular includes the discovery of the ozone hole. Several further TOMS instruments have been launched on other satellites. Like the Dobson instruments on the ground they yield very accurate O3 total column densities using a relatively simple method. Besides events of very strong atmospheric SO2 absorption and aerosols they are, however, only sensitive to O3.

    hdoas_fig10 (17K)

    Fig. 10 Overview on the history of satellites carrying UV/vis sensors.

    Since April 1995 the first DOAS instrument is operating from space. The Global Ozone Monitoring Experiment (GOME) was launched on the European research satellite ERS-2 [ESA, 1995; Burrows et al., 1999].

    Like SBUV and TOMS also GOME is a nadir viewing instrument; unlike its predecessor instruments it covers a large spectral range (240 - 790 nm) at a total of 4096 wavelengths arranged in four ‘channels’ with a spectral resolution between 0.2 and 0.4 nm. Its normal ground pixel size in 320 x 40 km²; the global coverage is achieved after three days.

    hdoas_fig12 (10K)

    Fig. 11 Similar to Dobson instruments (see Fig. 6) also the first O3 sensors from space measured the reflected solar intensity in specific narrow wavelength intervals.

    For O3 profile measurements the intensities at short wavelengths are observed (BUV/SBUB instruments); for the determination of the total atmospheric O3 column the intensities at larger wavelengths are used (TOMS instruments).
    In contrast to the limited spectral information of BUV/SBUV and TOMS instruments GOME spectra yield a surplus of spectral information (see Fig. 3 and 11). By applying the DOAS method to these measurements it is thus possible to retrieve a large variety of atmospheric trace gases, the majority of which are very weak absorbers (O3, NO2, BrO, OClO, HCHO, H2O, O2, O4, SO2).

    In addition also other quantities like aerosol absorptions, the ground albedo or indices characterising the solar cycle can be analysed. Because of the high sensitivity of GOME it is in particular possible to measure various tropospheric trace gases (NO2, BrO, HCHO, H2O, SO2) [Burrows et al., 1999; Wagner et al., 2002a]. A further important advantage is that the GOME spectra can be analysed with respect to a spectrum of direct sun light, which contains no atmospheric absorptions.

    Therefore in contrast to ground based DOAS measurements the DOAS analysis of GOME spectra yields total atmospheric column densities rather than the difference between two atmospheric spectra.

    hdoas_fig13 (20K)

    Fig. 12 Spectral coverage of the satellite DOAS instruments GOME and SCIAMACHY. From the IR measurements of SCIAMACHY also greenhouse gases and CO can be retrieved. (© Uni-Bremen, http://www.iup.physik.uni-bremen.de/sciamachy/)

    In March 2002 a second DOAS satellite instrument, the SCanning Imaging Absorption SpectroMeter for Atmospheric ChartographY (SCIAMACHY) [Burrows et al., 1988; Burrows et al., 1995; Bovensman et al., 1999] was launched on board of the European research satellite ENVISAT. In addition to GOME it measures over a wider wavelength range (240 nm - 2380) including also the absorption of several greenhouse gases (CO2, CH4, N2O) and CO in the IR (see Fig. 12).

    It also operates in additional viewing modes (nadir, limb, occultation), which allows to derive stratospheric trace gas profiles. As another advantage is that the ground pixel size for the nadir viewing mode was significantly reduced to 30 x 60 km² (in a special mode even to 15 x 30 km²). Especially for the observation of tropospheric trace gases this is very important because of the strong spatial gradients occurring for such species.

    The first tropospheric results of SCIAMACHY show that it is now possible to identify pollution plumes of individual cities or other big sources. Several additional space borne DOAS instruments are planned for future missions. Three instruments of a second generation (GOME-2) of GOME instruments scheduled planned for the EUMETSAT MetOp 1, 2, and 3 platforms (2005-2020), extending the GOME and SCIAMACHY atmospheric chemistry measurement series in the UV/VIS into the next two decades.

    In addition, the Ozone Monitoring Instrument (OMI) will be launched in 2004 and will further improve the spatial resolution with ground pixels of only 13x24km². Despite the various advantages of the surplus of spectral information the application of the relatively complex DOAS analysis to the spectra of GOME and SCIAMACHY also includes several important challenges. These challenges are further increased by several instrumental shortcomings, which especially complicate the retrieval of the weak atmospheric absorptions.

    A second large challenge addresses the interpretation of the slant atmospheric column densities derived from the DOAS fit of satellite spectra, which includes the numerical modelling of the atmospheric radiative transport.