University of Bremen IUP DOAS





DOAS Introduction
Airmass Factors
DOAS Glossary
Lecture Material
City of Science

DOAS Glossary

A   B   C   D   E   F   G   H   I   J   K   L   M   N   O   P   Q   R   S   T   U   V   W   X   Y   Z

Below you will find brief definitions and explanations of terms frequently used in the context of Differential Optical Absorption Spectroscopy (DOAS) measurements and retrievals. The discussion is strongly biased to scattered light measurements from the ground and from satellites, and some terms might have different meaning in other contexts. If you have comments or suggestions, please send a mail to Andreas Richter.

This page is still under construction and will be updated throughout the next weeks!

absorption cross-section  factor of proportionality between the number of molecules and the light absorption of a substance. The absorption cross-section depends on wavelength and in general on temperature and pressure. It is given in units of [cm2 per molecule].

aerosols solid or liquid particles suspended in the atmosphere. From the point of view of DOAS measurements, aerosols contribute to the extinction of the signal, both by scattering and by absorbing light. At the same time, they scatter light in the direction of the telescope and therefore also act as light source for scattered light observations. For MAXDOAS measurements, aerosols determine the length of the light path in the lower atmospheric layers and thereby the sensitivity of the measurements. At very large aerosol load, MAXDOAS measurements can loose most of their vertical resolution. For zenith-sky measurements, aerosols in the stratosphere can have a similar effect of reducing the light path and thereby the sensitivity. The largest problem however is not the change in light path itself which in principle can be accounted for but the uncertainty in how much it is changed. FOR MAXDOAS measurements, analysis of the absorption of the oxygen dimer O4 can be used to estimate the aerosol optical depth and a to derive a rough vertical aerosol profile. By using measurements in different directions, also information on the aerosol phase function and thereby the aerosol composition can be derived.   

AMAXDOAS (airborne multi-axis DOAS) Airborne DOAS instrument observing scattered light under different elevation angles. By combining nadir and zenith measurements, the absorber columns above and below the aircraft can be separated. For an aircraft flying at tropopause altitude, this facilitates separation of stratospheric and stratospheric columns. By using viewing directions pointing close to the horizon, vertical information can be derived close to the flight altitude similar to limb measurements. If measurements at different wavelengths are combined, the vertical resolution improves.

AMF (airmass factor) Factor of proportionality between the slant column which is retrieved from the DOAS fit and represents the integrated amount of absorber averaged over all light paths and the vertical column which is defined as the column of the absorber integrated vertically through the atmosphere or a given layer. Airmass factors are sometimes also called enhancement factors and can be interpreted as indicating the enhancement in sensitivity or light path relative to the vertical transect. Airmass factors depend on geometry (solar zenith angle, viewing direction, measurement altitude, ...), atmospheric composition (vertical profile of absorbers and aerosols, temperature profile, ...) and boundary conditions (surface albedo, clouds, ...). Airmass factors are calculated using radiative transfer models such as SCIATRAN. For stratospheric absorbers, not too low sun (SZA < 70) and weak absorption, the geometric approximation of 1 / cos(SZA)  for ground-based and (1 / cos(SZA) + 1 / cos(LOS) for satellite measurements is a good first approximation.

averaging kernel The averaging kernel describes how a quantity retrieved from a remote sensing measurement changes in response to a change in one of the parameters. In the context of satellite DOAS measurements, the column averaging kernel of e.g. NO2 describes how the retrieved vertical column changes in response to a change of NO2 concentration at a given altitude. It is important to realise that the averaging kernel depends both on the measurement sensitivity described by the weighting function and the assumptions made for the retrieval. It therefore can be used to remove the assumptions made in the retrieval on the vertical distribution of the absorber and to make a consistent comparison between a column predicted by a model and the satellite measurement.

background spectrum For an absorption measurement based on Lambert-Beer's law, a reference or background intensity I0 is needed. Ideally, this measurement should have no absorption of the absorber of interest or - if that is not possible - at least a small or well known amount. For satellite measurements, direct solar observations are often available and can be used as background spectrum. For ground-based measurements, no absorption free background is available and usually a measurement taken at high sun in zenith-direction is used as background. To account for the residual absorption in this spectrum, either a Langley-plot is used or other information (e.g. an ozone sonde measurement taken on the same day) is used to estimate it. As the background spectrum is mainly used to remove the Fraunhofer structures from the spectra, it sometimes is also called Fraunhofer background. In long path DOAS measurements where a lamp with a smooth spectrum is used as light source, the initial intensity can often be approximated by a polynomial and no background spectrum is needed.

chemical airmass factors If an absorber undergoes rapid photochemistry (e.g. OClO), its concentrations change during twilight. The observed change in slant column is therefore determined by the combination of photochemistry and change in light path. The situation is further complicated by the fact, that the local solar zenith angle (which determines the photochemical equilibrium) changes with altitude along the light path of the scattered photons. All these effects can be simulated by combining the output of a chemical model with the radiative transfer model when computing airmass factors, and the resulting values are sometimes called chemical airmass factors. 

block airmass factor The airmass factor (AMF) is defined as the ratio of slant column and vertical column. Usually, the AMF is computed for a standard vertical profile of the absorber. To study the change of sensitivity with altitude, airmass factors are computed for thin layers in the atmosphere and displayed as  function of height. These altitude dependent airmass factors are called block airmass factors. If the atmosphere is optically thin with respect to the absorber of interest, any airmass factor can be computed as a weighted mean of the product of the block airmass factors with the vertical profile of the absorber. The block airmass factor is identical to the relative weighting function of the slant column. It is also related to the averaging kernel of a measurement.

CCD Charged Coupled Devices are often used as detectors in DOAS instruments. The main advantages of these detectors are their very low dark signal, in particular if cooled, and the option to retrieve the 2-dimensional distribution of the intensity. Low dark signal facilitates measurements at low intensities and improves signal to noise. The 2d-capabilities can be employed in imaging spectrometers to take measurements in different directions simultaneously. The disadvantages of CCDs are their relatively small full well capacity which necessitates frequent read-out of the detector and the long time it takes to read out the detector, in particular in 2d mode. Both aspects have improved over the last years e.g. with frame-transfer type cameras.

cloud fraction At the spatial resolution of current satellite instruments used for DOAS measurements, many measurements are over partially cloud covered scenes. As clouds are very much brighter than most surfaces (the exception being snow and ice), even a small cloud can have a large effect on the signal measured from satellite. As an example, if we assume the cloud albedo to be 80% while the surface albedo is typically 5% in the UV, a cloud covering 10% of a ground scene will reflect 0.8 * 0.1 =   0.08 of the incoming radiation while the surface reflects only 0.05 * 0.9 = 0.045 of the incoming photons. Therefore, in first approximation nearly two thirds of the photons collected by the satellite instrument originate from the cloud even at only 10% geometrical cloud cover in this simplified example. In a more realistic calculation, Rayleigh scattering in the atmosphere needs to be accounted for which will reduce the impact of the cloud as well as varying optical thickness of clouds which results in lower albedo for thinner clouds..

clouds The presence of clouds has a strong impact on the radiative transfer in the atmosphere and thereby on the light path in scattered light observations. Depending on the viewing geometry and the vertical profile of the absorber of interest, the effect of clouds can be very different. For zenith-sky observations of stratospheric absorbers, clouds have a minor impact as they mainly change the light path in the lower troposphere and not in the stratosphere. This is one of the advantages of zenith-sky observations which provide data under almost all weather conditions. In the case of very high clouds such as PSCs or strong volcanic aerosols, the light path in the stratosphere is also changed and this has to be taken into account. For ground-based measurements of trace gases which are present in the troposphere (e.g. NO2), light path enhancement within the cloud can greatly increase the observed signal depending on the optical thickness of the cloud and the absorber vertical profile. Horizon pointing measurements are affected by clouds in different ways - high clouds have little impact as the bulk of the absorption takes place in the slant path in the lowest atmospheric layers while lower clouds can reduce the achievable path length in a given viewing direction. In the presence of very low clouds (fog), the observed signal is the same in all directions. Satellite measurements are affected by the presence of clouds in mainly two ways: The effective albedo increases (albedo effect) which improves the signal to noise ratio and at the same time improves the sensitivity of the measurements above the clouds. On the other hand, the part of the atmosphere below the cloud is effectively hidden from the measurement (shielding effect, see also ghost column). Within the uppermost part of the cloud, multiple scattering can increase the sensitivity for satellite measurements but compared to ground-based measurements, the effect is small.

column Integrated amount of an absorber, usually given as column density (number of molecules per cm2). The vertical column is integrated vertically and therefore independent of the measurement geometry while the slant column is integrated along the light path through the atmosphere. For scattered light observations, many light paths contribute and the slant column is a weighted average over all contributions. Conversion between slant and vertical column is performed by applying airmass factors from radiative transfer calculations.

dark current Current observed in a diode array or CCD detector without external light source. It originates from thermally induced electron hole pairs and therefore depends exponentially on temperature. As a rule of thumb, cooling the detector by 7K will reduce dark current by a factor of two. As the signal, dark current is proportional to the integration time, and thus becomes relevant for low light levels and long integration times. When cooled to -40C or lower, dark current of scientific grade CCDs is very small for typical integration times. As a complication, dark current in diode arrays depends on the amount of charge already collected, and therefore has a non-linear component. Usually, one tries to minimise dark current in order to use as much of the available dynamic range of the detector as possible but also because the shot noise of the dark current contributes to the overall noise of the measurement.

dark signal Total signal observed in a measurement without light source. It includes the dark current, but usually also an electronic offset and in the case of IR measurements the contribution of thermal emission of the instrument itself. As it is composed of different contributions, it is not necessarily proportional to the integration time used. In complex instruments such as GOME or SCIAMACHY, the dark signal of one detector can depend on the readout of another detector through electronic cross-talk. Therefore, dark signals are measured by GOME with exactly the same settings as the normal measurements only on the dark side of the orbit.

dichroic filter is used in the GOME and SCIAMACHY instruments to separate the incoming lights into the different spectral channels. The dichroics used have a throughput that rapidly varies with wavelength close to the cut-off wavelength, the details of which depend on the polarisation of the incoming light. Therefore, only very accurate calibration removes residual structures from the spectra taken by  GOME and SCIAMACHY in the affected spectral regions. In particular for GOME, the dichroic characteristics changed between the pre-launch calibration measurements which were in part done at ambient pressure and humidity and the in-orbit situation. Therefore, large (up to the percent level) residual structures remain in the calibrated spectra of GOME in certain spectral regions which interfere with the retrieval of trace gases using the DOAS method.

diffuser plate Satellite instruments such as GOME, GOME-2 and SCIAMACHY take measurements of scattered light and once per day also a direct sun measurement as absorption free background spectrum. The latter has a much higher intensity, and to avoid detector saturation, the intensity must be reduced either by internal stops or by use of a diffuser plate. Both GOME and SCIAMACHY use a diffuser plate for solar irradiance measurements which provides good filling of the instrument slit. However, in the case of GOME, the diffuser plate introduces spectral artefacts into the measurements which depend on the angle of illumination. The origin of these structures is interference on the semi-periodic structures created on the diffuser during sand blasting. As the angle between satellite and sun changes during the year, the structures also change and introduce systematic and reproducible offsets in trace gas columns of weaker absorbers if daily solar spectra are used in the data analysis. These offsets have to be corrected by using external information, e.g. from validation experiments. Fortunately, a similar effect is not observed with the SCIAMACHY and GOME-2 diffuser plates.

Doppler shift For the DOAS retrieval, an absorption free background spectrum is needed. Satellite sensors which are outside the atmosphere use direct sun observations to obtain this reference spectrum, usually via a diffuser plate. Both GOME and SCIAMACHY take the solar measurement in flight direction, and therefore the measurements experience a Doppler shift as result of the satellite speed of approx. 7000 m/s. This results in a 0.008 nm shift at 350 nm. As the nadir measurements are taken orthogonally to flight direction, they are not Doppler shifted and before taking the ratio of the two measurements, the wavelength shift must be corrected.

etalon Interference pattern observed on linear detectors through multipath interference of the incoming light in passivating layers or ice formed on the detectors. Depending on the refractive index and the thickness of the layers as well as the angle of inclination, the period of the observed intensity variations can vary but often is longer than that of the absorption structures studied. If the pattern is stable with time, it cancels when taking the ratio with the background spectrum, at least if no wavelength shift has to be applied. However, if it is changing with time, then high frequency structures show up in the ratio which can strongly interfere with the DOAS retrieval. In the GOME instrument, changing etalons are observed in all channels whenever the detector cooling had to be switched off temporarily. The effect stabilises after a few days as the amount of water vapour in the instrument decreases. To reduce etaloning, several approaches can be taken, most importantly preventing ice formation by evacuation of the detector, removal of quartz covers or replacement by wedged windows.

filtering In the DOAS retrieval, two forms of filtering are often applied: high pass filtering for the separation of the structured absorptions from trace gases from the background extinction by air and aerosols and low pass filtering for the reduction of noise. Low pass filtering is an essential part of the DOAS approach and is often accomplished by introducing a wavelength dependent polynomial of low order (3 - 5 coefficients) into the retrieval but other filtering approaches can also be used (using the derivative or Fourier filtering). High pass filtering on the other hand is often not used at all; if the signal to noise ratio of a measurement is not high enough, several measurements are averaged.

Fraunhofer lines Absorption lines in the solar spectrum. Fraunhofer lines affect scattered light DOAS in three ways: i) The wavelength adjustment of the spectra must be very good to avoid artefacts from incomplete removal of Fraunhofer lines when taking the ratio of measured and background spectrum; (ii) In scattered light, inelasitc scattering (e.g. rotational or vibrational Raman scattering) reduces the apparent depth of Fraunhofer lines depending on the relative contribution of inelastically scattered photons. This has to be accounted for or incomplete removal of Fraunhofer lines leads to large residuals; (iii) Usually, it is assumed that absorption in the atmosphere and convolution of the spectra by the instrument slit function can be exchanged in the analysis so that the absorption cross-sections can directly be convoluted with the instrument slit function prior to the DOAS fit. As result of correlation of the Fraunhofer lines with absorption lines this does not really hold for species with highly structured absorption spectra, leading to the so called I0-effect.

ghost column In UV/vis satellite measurements, clouds limit the ability of the instrument to probe the atmosphere down to the surface. In first approximation, the column retrieved in a cloudy situation corresponds only to the column above the cloud although thin or broken clouds may permit some signal from the atmosphere below the cloud to reach the satellite. In order to determine the total atmospheric column for such cases, the hidden "ghost" column below the cloud has to be added to the measured column. This is usually done by either adding a value from a climatology or model or by assuming that the profile used in the airmass factor calculation is correct and applying it to a cloudy situation. The latter approach essentially scales the a priori profile with the visible part. In situations with broken clouds, two numbers are usually calculated, one for the cloud free part of the pixel and one for the cloudy part and the results weighted by cloud fraction. While addition of a ghost column improves the consistency of a data set in particular in cases where most of the absorber are located above the cloud (e.g. ozone) it can introduce large errors for individual measurements of absorber which are mainly in the lower atmosphere (e.g. NO2).

GOME The Global Ozone Monitoring Experiment is a UV/visible grating spectrometer observing the Earth in nadir viewing geometry from the European ERS-2 satellite. GOME has 4 channels, each of which is equipped with a grating and a 1024 pixel photo diode array. The spectral coverage of GOME is 237 - 793 nm at a resolution of about 0.2 nm FWHM below 400 nm and 0.4 nm at longer wavelengths. With a swath width of 960 km which is divided into three forward scans of 320 x 40 km2 each and a rapid backward scan it achieves global coverage at the equator within 3 days. The ERS-2 satellite was launched into a sun-synchronous orbit with a descending node equator crossing time of 10:30 LT in April 1995 and GOME nadir data are available from July 1995 onwards. As result of the failure of the last on-board tape recorder in June 2003, GOME coverage is limited to regions with direct data downlink since then. More on the GOME instrument can be found on our GOME page. The GOME instrument can be seen as predecessor of both the SCIAMACHY and GOME-2 instruments.

GOME-2 To extend the GOME and SCIAMACHY time series, EUMETSAT operates a series of three  GOME-2 on the MetOp platforms. The first GOME-2 instrument was launched in October 2006 and is providing data since March 2007. It is expected that the data from the three GOME-2 instruments will eventually cover 15 years. The GOME-2 instrument is very similar to GOME but has improved spatial resolution (40 x 80 km2 over most of the orbit) and has a much wider swath (1920 km) which results in global coverage at the equator after 1.5 days. Additional improvements are a much higher spectral resolution of the onboard polarisation measurement devices (PMDs) which in combination with the reduced sensitivity to polarisation significantly reduce the impact of instrument features on the spectra measured. As ERS-2 and ENVISAT, MetOp is in a sun-synchronous polar orbit; the equator crossing time (descending node) is 09:30 LT.

grating Diffractive element used in most spectrometers employed for DOAS applications. Compared to prisms, gratings offer better spectral resolution. As grating efficiency is polarisation dependent, instruments used for scattered light measurements must either use a polarizer in front of the slit or efficiently depolarise the incoming light to avoid problems when comparing measurements taken at different solar azimuth angles. For the GOME and SCIAMACHY instruments, the polarisation dependency was measured before launch and is corrected for each spectrum based on the signals of the on-board polarisation measurement devices (PMD). However, residual structures remain as result of calibration errors and instrumental changes, leading to interference problems in the DOAS retrieval. This problem has been solved for OMI by using a polarisation scrambler.  


imaging DOAS DOAS instrument that creates a 2-dimensional "picture" of the absorber columns. This is usually achieved by combining an imaging spectrometer having a 2-dimensinal detector with a sequential scan (e.g. OMI), but scans in both directions are also sometimes used together with a 1-dimensional detector (e.g. GOME and SCIAMACHY). For satellite measurements, imaging DOAS is the standard approach as they aim at providing global maps. For ground-based instrumentation, imaging DOAS is mainly used for monitoring of strong, localised sources such as chimneys and volcanoes.

imaging spectrometer Spectrometer that creates an image of the slit on the detector, facilitating separation of signals from different parts of the slit. By using a 2-dimensional detector (usually a CCD) that maps wavelength on one axis and the position on the slit in the other, many spectra can be obtained at the same time.  

Lambert-Beer's law The change in intensity as a function of distance the light travels in a medium such as the atmosphere is described by Lambert-Beer's law, which sometimes is also called Bouguer's law. It states that the change in intensity (dI) over an infinitesimal path (ds) is proportional to the concentration of the absorber (c) and its absorption cross-section (s): dI / ds = -c s. By integrating over the light path, the total absorption can be determined. As the absorption cross-section is a function of wavelength, this has to be taken into account. Lambert -Beer's law can easily be extended to include several absorbers and extinction by scattering.

Langley plot Graph which originally was used to determine the extraterrestrial radiance at one wavelength from gorund-based measurements. By taking measurements at many different solar elevations and plotting intensity as a function of 1 / cos(SZA), a straight line is found which can be extrapolated to SZA = 0 or no air mass. Applied to measurements of slant columns, the relation SC = VC * AMF(SZA) is used to plot the measured SC as a function of AMF which should result in a straight line with the slope of VC and an offset of SC0, the amount of absorber in the background spectrum.

limb measurements observe scattered or emitted light in directions close to the horizon, usually from a balloon, aircraft or satellite. Similar to occultation measurements, most of the signal originates from the area around the tangent point (the point closest to the Earth's surface). By scanning the atmosphere in different elevation angles, vertical profiles of atmospheric absorbers (or emitters) can be retrieved, the weighting functions having pronounced peaks in one altitude. In contrast to occultation measurements, limb measurements can provide global coverage. The disadvantage is, that radiative transfer calculations are more complex and as result of increased scattering, limb measurements can usually not probe the troposphere. The satellite instrument SCIAMACHY alternates limb and nadir measurements, combining the advantages of both viewing modes at the expense of spatial coverage. Limb measurements are also used by the AMAXDOAS instrument.

limb darkening

long-path DOAS Instrument using a strong light source for an open path absorption measurement in the troposphere. Often the lamp is situated close to the spectrometer and directed to an array of retro-reflectors from which it is picked up by a telescope. Long-path DOAS measurements integrate over a larger volume than in-situ measurements but over a shorter distance than scattered light observations. Depending on the length of the light path and the absorption cross-section of the species of interest, the detection limits of the method can be very low. In contrast to scattered light measurements, UV light is not absorbed in the ozone layer which extends the range of observable species to many organic compounds. Also, night-time measurements are possible. By combining measurements at different altitudes or scanning reflectors mounted at different altitudes, vertical information can also be retrieved from long-path measurements.

LOS (line of sight) For satellite measurements, the angle between nadir and the viewing direction is called line of sight angle. For GOME and SCIAMACHY, it varies between 0 and 34.

MAXDOAS Multi Axis Differential Optical Absorption Spectroscopy uses measurements in the zenith and in several viewing directions close to the horizon to separate tropospheric and stratospheric absorptions and to retrieve profiles in the lower troposphere. The measurements are either taken sequentially by rotating the whole telescope or using a mirror or simultaneously by employing an imaging spectrometer and an array of small telescopes pointing in different directions. More on the MAXDOAS technique can be found on our MAXDOAS page.

nadir measurements Downward looking measurements from an aircraft or satellite. Depending on the instrument, the viewing direction is usually varied across track either by a scanning mirror or by use of an imaging spectrometer. As long as the instrument is still viewing the surface, the measurement is considered to be nadir; if not, it is a limb measurement. Nadir measurements provide column information unless additional information can be used to derive vertical resolution. Examples are line widths in the IR, wavelength dependent penetration depths for strong absorbers such as O3 in the UV or CO2 in the IR, or temperature dependent absorptions / emissions in combination with information on the atmospheric temperature profile. DOAS nadir measurements are performed from satellites (GOME, SCIAMACHY, OMI) and from the airborne AMAXDOAS instrument.

narrow swath mode On some days (the 5th, the 10th, and the 25th of a month) the GOME instrument is operated in a special "narrow swath" mode. In this mode, the spatial resolution is improved to 80 x 40 km2 but the spatial coverage is reduced as result of the smaller swath (240 km instead of 960 km). A similar mode is sometimes used from GOME-2 where spatial resolution is then improved to 20 x 40 km2 and the swath is reduced from 1920 km to 480 km.

O4 oxygen dimer is formed in the atmosphere by collision of oxygen molecules and can easily be detected by DOAS measurements. O4 has several strong bands throughout the UV and visible wavelength range which are only weakly dependent on temperature. As O2 is well mixed in the atmosphere, its vertical profile can be computed from surface pressure and thus also the vertical profile of O4. Measurements of O4 can therefore be used to determine the average length of the light path through the atmosphere which can be employed to iteratively improve cloud and aerosol settings in the radiative transfer calculations needed for the determination of airmass factors. Potentially, this method can also be used to retrieve aerosol optical depth and by combining measurements at different wavelengths, viewing directions and absolute intensities also of aerosol type and vertical profile.

OMI Ozone Monitoring Instrument. Nadir viewing imaging spectrometer covering the wavelength region from 270 to 500 nm at 0.5 nm resolution. By using a CCD as a detector, the across track swath is imaged without a scanning mirror providing ground pixels from 13 x 24 km2 to 120 x 24 km2 depending on their position within the swath. OMI was launched on AURA in July 2004 and has a 2600 km swath which provides daily global coverage. More on OMI can be found on the OMI Homepage.

orbit Track of a satellite around the Earth. For Earth observation, low earth orbits (LEO) and geostationary orbits (GEO) are relevant. In LEO, the satellite circles the Earth, in the case of ENVISAT and ERS-2 14 times per day. These two platforms are in sun-synchronous near polar orbits, circling around the Earth at about 850 km altitude crossing each latitude band at a fixed local time. Depending on the swath width of the instruments, measurements from LEO provide global coverage after one or several days but with the exception of high latitude regions can not sample the diurnal variation of atmospheric composition. Geostationary satellites in contrast are in an orbit around the equator at about 36,000 km altitude where they have just the right speed to appear stationary relative to the Earth's surface. From this position, they can continuously observe a part of the Earth providing several measurements per day but no global coverage. In particular, they can not cover the polar regions.

off-axis measurements Scattered light measurements pointed not to the zenith but to the horizon. These measurements were first used in stratospheric research to improve the signal in twilight measurements by looking at the bright part of the sky, but are now mainly used for tropospheric measurements. The basic idea is, that the viewing directions with small elevation angles have a very long light path in the lowest atmospheric layers while the light path in the upper troposphere and stratosphere does not depend strongly on the viewing direction. More on the MAXDOAS technique can be found on our MAXDOAS page.

PDA photo diode array

photochemical correction At twilight, the concentrations of photochemically active species vary with solar zenith angle, in some cases (e.g. OClO) quite dramatically. As the solar zenith angle varies along the light path through the atmosphere, a measurement at twilight probes different photochemical regimes in different altitudes. Therefore, the conversion of the measured slant column to a vertical column is not straight forward and comparison of modelled and measured slant columns is more appropriate. This can be achieved by using a chemical model to determine the dependence of e.g. OClO concentration on solar zenith angle for each altitude, and introducing this dependency in the radiative transfer code. The result is a photochemically corrected modelled slant column that can be directly compared with the measurements.

PMD Polarisation Measurement Device of the GOME, GOME-2 and SCIAMACHY instruments. The PMDs are used to measure the polarisation state of the atmosphere for each measurement, an information needed to correct for the polarisation sensitivity of the instruments. The PMDs are broad band semi conductor detectors that are read out at higher frequency than the diode arrays of the instruments and therefore provide superior spatial coverage. They are therefore also used to determine cloud fraction by looking at intensity, "whiteness" and intensity variability of the measurements.

polarized measurements Most DOAS measurements are not set-up to measure only one polarisation direction but rather aim at taking a polarisation independent measurement, e.g. by using a quartz fibre bundle. However, polarisation is relevant for scattered light measurements for two reasons. First, many grating spectrometers (and all instruments using mirrors) have polarisation dependent throughput which has to be corrected to avoid interference with the retrieval of weak absorption signals or reduced by depolarising the light before it enters the spectrograph. The second effect of polarisation is that different scattering processes in the atmosphere (Rayleigh, Raman, Mie) have different effects on the polarisation state of the scattered light, and by selecting a specific polarisation direction, one can select or reject photons that have undergone e.g. Rayleigh scattering (see also Ring effect). Therefore, some instruments use a polarisation filter in front of the entrance optics and track the solar azimuth over the day to ensure that they observe mainly Rayleigh scattered photons under well defined scattering geometry to simplify radiative transfer calculations.

polynomial fits are used in DOAS applications for two purposes. First, they remove the varying contribution of scattering to the extinction of the signal as both Rayleigh and Mie scattering have a wavelength dependence that can be approximated by a polynomial of low order. The second reason for using a polynomial in the DOAS fit is the separation of low frequency and high frequency parts of the absorption cross sections which helps to separate different absorbers in the spectra. The latter can also be achieved by using high pass filtered data or the first derivative of the absorption cross-sections. In practice, the polynomial fit can also compensate some instrumental problems such as etalons.

quartz fibre bundle Ground-based DOAS instruments often use quartz fibre bundles to connect the spectrometer with the telescope. This facilitates flexible set-up of the instruments, for example with the spectrometer in a temperature controlled room and the telescope on the roof of the building. In addition, the quartz fibre bundle can be built to be circular on the telescope end and forming a slit on the spectrometer end, thereby optimising instrument throughput. Another benefit from using quartz fibre bundles is the fact, that light is efficiently depolarised when leaving the bundle. This is important as sky light is polarized (Rayleigh scattering) and many grating spectrometers are sensitive to the polarisation of the incoming light which complicates comparison of measurements taken under different solar positions.

radiative transfer model In order to simulate the radiance received by an instrument measuring in or above the atmosphere, radiative transfer models are used. These software codes account for all or part of the processes in the atmosphere, including scattering on molecules and aerosols, absorption by trace species and aerosols, reflection on the surface, refraction, the geometry of sun, atmosphere and instrument, etc. They can either be used as forward models to simulate the expected radiance for a given atmospheric situation or iteratively in a retrieval by varying the atmospheric parameters until satisfying agreement is found with the measurements. For the use in a retrieval, it is advantageous to have the weighting functions also computed by the model. There are mainly two different approaches used for radiative transfer models. In the first approach, the physics of radiative transfer are directly simulated by following the fate of a large number of photons (see ray tracing). This approach is based on first principles, can relatively easily incorporate any desired physical effect in three dimensions and can be used to investigate photon statistics. However, it is quite computing time intensive if high accuracy is required.

ray tracing

retro reflector

Ring effect

rotational Raman scattering


shift and squeeze

slant column


slit function

solar zenith angle (SZA)

southern Atlantic anomaly (SAA)

state SCIAMACHY measurements are divided into orbits (one circling around the Earth) and states, which are one set of measurements taken with the same instrument settings (instrument viewing mode and integration time).



twilight measurements In the zenith-sky viewing geometry, the light path through the stratosphere is maximised at twilight. This is used to get a large signal even for very weak absorbers. The problem is, that at low sun also the intensity of the scattered light decreases which reduces the signal to noise ratio and necessitates longer integration times (several minutes instead of seconds at noon). Another problem is that photochemically active species such as NO2, BrO, or OClO change their concentrations at twilight, either by their own photolysis or by photolysis of their precursors and reservoir species. As a result, the vertical column of these species can not be considered to be constant throughout one measurement. This can be accounted for by applying a photochemical correction to the airmass factors. 


vertical column

weighting function

vibrational Raman scattering

wavelength calibration The raw measurements taken by a spectrometer are still a function of pixel number which has to be converted to wavelength in nm. This is usually done by using a light source with emission lines at well known wavelengths, e.g. a mercury line lamp. In a typical application, several lines are found within the wavelength region observed and a polynomial is fitted through the points which is then used to compute the wavelength values for all pixels of the detector. While this procedure is accurate to better than 0.1 nm, a wavelength accuracy of about 0.01 nm is needed for reliable retrieval of weak absorbers such as BrO. This is achieved by fitting the measurement on a direct sun measurement taken at high spectral resolution and convoluted to the resolution of the instrument. As there is a very large number of  Fraunhofer lines in the UV and visible part of the spectrum, this correlation is more accurate than the initial calibration using just a few lines. If an analytical expression for the slit function is known (e.g. a Gaussian), the slit function parameters can simultaneously be determined in the fit.

zenith-sky measurements Originally, ground-based measurements of scattered light were aimed at the observation of stratospheric absorbers. Therefore, the instrument telescope was directed towards the zenith-sky as in this viewing geometry the tropospheric light path is minimised while the stratospheric light path is maximised. This is particularly true during twilight as illustrated on our MAXDOAS page. For measurements of tropospheric species the zenith-sky geometry provides low sensitivity and therefore usually is employed to provide the background measurement.

web master:      Last Update: 11-04-16