Hyperspectral remote sensing is a challenge to data analysis due to the large data rate. Since it will often not by possible to analyze the entity of measured spectral data, so-called 'micro windows' are selected which contain the bulk of information on the target state parameters to be retrieved, while any interfering signal is minimized. We discuss the benefits of a quantitative method for automatic selection of optimized sets of such micro windows for the analysis of Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) data. While MIPAS, which will be operated onboard the polar orbiting environmental satellite (ENVISAT), is not a hyperspectral instrument, it serves as an example to study the power of the micro window approach. The MIPAS instrument will measure the IR emission of various atmospheric trace gases by limb scans covering the altitude region from 6 to 68 km altitude.

The Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) will be operated onboard the polar orbiting Environmental Research Satellite (ENVISAT). The instrument will measure the infrared emission of various atmospheric trace gases by limb scans covering the altitude region from 6 to 68 km altitude.Infrared spectra are usually not evaluated as a whole, but in small spectral regions, where the target gases have prominent, but nonsaturated transitions and the contributions of non-target species are low. We present the results of a quantitative method for automatic selection of optimized sets of these so-called "microwindows" for the analysis of MIPAS data(1,2). For each target species and tangent altitude this method identifies all spectral regions, where the total retrieval error exhibits a local minimum. For this purpose random noise and various uncertainties in atmospheric and instrumental parameters are taken into account.Microwindow databases and optimum subsets, so-called "occupation matrices", were calculated for midlatitude and polar winter conditions for the main target quantities temperature, line-of-sight, CH4, H2O, N2O, O-3, HNO3 and NO2, but also for various minor constituents by using synthetic MIPAS-spectra calculated in Oxford and in Karlsruhe, respectively. After a short descrition of the theory and of the optimization algorithm we will compare midlatitude and polar retrieval error profiles. Further we will estimate the suitability of the midlatitude occupation matrices for evaluation of polar measurements and vice versa by inspection of the change in retrieval error.

Test retrievals for the airborne limb- and upward sounder MIPAS-STR (Michelson Interferometer for Passive Atmospheric Sounding-STRatospheric aircraft) were performed. The influence of spectral noise and systematic errors was tested for a set of atmospheric trace gases. Downward error propagation from the unknown part of the profile above the aircraft and unresolved features in the reference profiles lead to deviations of the fit result even for the unperturbed retrievals. In cases of smooth profiles it was possible to retrieve the gradient up to some km above the Right level. Averaging kernels and test retrievals showed the possibility to increase the vertical resolution by oversampled measurements. However, a field-of-view (FOV) oversampling with respect to a vertical FOV extent larger than 1/2 does not seem to improve the results significantly.

The possibility to derive microphysical properties of polar stratospheric clouds from future MIPAS (Michelson Interferometer for Passive Atmospheric Sounding) -ENVTSAT measurements was investigated. Available refractive index data for PSC candidates were intercompared in order to estimate their reliability. Especially for NAT the laboratory measurements differ significantly and for ternary H2SO4/HNO3/H2O solutions only one source of data exists. For simulating limb-spectra, a Mie model was implemented in the forward code KOPRA (Karlsruhe Optimized and Precise Radiative transfer Algorithm) in such a way that in parallel to the radiance spectra the derivatives with respect to a variety of microphysical. aerosol parameters can be generated, Broadband forward calculations for small and large aerosols were made for various refractive indices. For Large particles the PSC signal in the spectrum was up to forty times larger than the noise level. The signal for small particles was around the spectral noise. By minimizing the total retrieval error an automatic microwindow selection was performed for different PSC scenarios. Under the assumption of known radius and width of the aerosol size distribution resulting errors for number density retrieval were less than 3% (9x10(-4) cm(-3)) for large and around 50% (7 cm(-3)) for small particles. For large particles it is possible to perform a two-parameter fit of number density and mode radius with errors less than 10%. Due to the Rayleigh-limit a distinction between radius and number density is not possible for small particles. However, the volume density can be derived with 12% (0.4 mum(3)/cm(-3)) uncertainty for large and 20% (2.2x10(-2) mum(3)/cm(-3)) for small aerosols. These conclusions are valid as long as the aerosol layer is not optically thick which in our examples was the case for water ice (type II) PSCs of number density 0.2 cm(-3) and radius 2.6 mum.

For data analysis of the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) atmospheric limb emission spectroscopic experiment on Environmental Satellite microwindows, i.e., small spectral regions for data analysis, have been defined and optimized. A novel optimization scheme has been developed for this purpose that adjusts microwindow boundaries such that the total retrieval error with respect to measurement noise, parameter uncertainties, and systematic errors is minimized. Dedicated databases that contain optimized microwindows for retrieval of vertical profiles of pressure and temperature, H2O, O3, HNO3, CH4, N2O, and NO2 have been generated. Furthermore, a tool for optimal selection of subsets of predefined microwindows for specific retrieval situations has been provided. This tool can be used further for estimating total retrieval errors for a selected microwindow subset. It has been shown by use of this tool that an altitude-dependent definition of microwindows is superior to an altitude-independent definition. For computational efficiency a dedicated microwindow-related list of spectral lines has been defined that contains only those spectral lines that are of relevance for MIPAS limb sounding observations.

The Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) [1] on board of the European Space Agency’s (ESA) Environmental Satellite ENVISAT-1 will be lauched into a polar orbit of about 800 km altitude.

We discuss the intercomparison between the Karlsruhe Optimized and Precise Radiative transfer Algorithm (KO-PRA) and the Reference Forward Model (RFM) codes, which have been designed for analysis of MIPAS-ENVISAT data. The purpose of this intercomparison is to validate the KOPRA algorithm, i.e. to identify and to remove possible errors in the KOPRA (or RFM) code and to quantify the reason of remaining differences. Similar comparisons between the MIPAS Optimized Forward Model (OFM) and the RFM as well as between KOPRA and the RFM have already been performed. To be able to relate on these results, this validation is similarly organized: we perform subsequently more complex tests of ray-tracing, integrated column amounts, homogeneous and limb path calculations of unapodised, apodised and field-of-view (FOV) convolved spectra, using the same isolated CO2 line as well as the same six MIPAS microwindows. Additionally we compare modeling of CO2 line-mixing, non-local thermodynamic equilibrium (NLTE), trace gas continua and cross-section spectra. The KOPRA-RFM residuals are below a quarter of the noise- equivalent spectral radiance (NESR) for the isolated CO2 line as well as for the MIPAS microwindows, i.e. KOPRA fulfills the acceptance criteria requested for the OFM. In most cases the deviations are even clearly below 1 nW/(cm2 sr cm-1), that is more than one order of magnitude below the acceptance threshold. This is valid for unconvolved as well as for ALS (apodised line shape) and FOV convolved spectra. There is also good agreement in modeling of the H2O-, O2- and N2-continua and of CO2 line-mixing. Larger deviations of up to several nW/(cm2 sr cm-1) occurred for NLTE calculations on the basis of 'default' atmospheric profiles with vertical resolution of 1 or 2.5 km. These differences were found to be due to different layer-averaging of the vibrational temperatures and could be considerably reduced by calculations with a higher vertical resolution of 250 m. Cross-section spectra agree well, if the tabulated data are given independent of pressure, e.g. for ClONO2 and N2O5, and cover the atmospheric temperatures. Due to different temperature extrapolation the deviations increase up to 10 nW/(cm2 sr cm-1) for atmospheric temperatures outside the measuring range. The RFM is not yet adjusted to cross-sections tabulated for non- equidistant temperatures and for atmospheric pressures, like CFC-data in the HITRAN96 database. If these data are used, larger differences arise, e.g. up to 30 nW/(cm2 sr cm-1) between CFC-12 spectra. Avoidance of interpolation by performing homogeneous path calculations for p,T of one of the tabulated cross-section datasets reduces the deviations to below 0.5 nW/(cm2 sr cm-1).

In atmospheric Fourier transform spectroscopy so-called microwindows are usually analyzed for retrieval of trace constituents rather than the spectrum as a whole. These microwindows, which are sets of consecutive spectral grid points, contain one or more prominent transitions of the target species, whereas it is desirable for the signal of interfering species to be minimum. An objective, quantitative method is presented to optimize the microwindow boundaries with respect to random errors, signal of interfering species, other parameter and systematic errors and to select optimum microwindows with respect to their associated retrieval errors. Case studies for N(2)O microwindows are performed for a spaceborne limb emission experiment to assess the dependence of the optimum microwindow width on the retrieval concept.

We present the Karlsruhe Optimized and Precise Radiative transfer Algorithm (KOPRA) which has been specifically da eloped for data analysis of the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) going to be launched on ESA's polar-orbiting Environmental Satellite 1 (ENVISAT-1) in 1999. KOPRA has been designed to account for the particular instrument requirements of MIPAS and the observation scenarii during the ENVISAT mission, in particular with respect to the viewing direction and the altitude coverage of the atmosphere. The conceptual details of KOPRA, which reflect the requirements set up by the instrument design details, the observation scenarii, and the link to a retrieval concept with high flexibility, are presented. The forward model error due to discarding individual physical processes and properties of the atmosphere as well as an over-all error budget with respect to these parameters is assessed in order to demonstrate the improvements of retrieval accuracy expected by usage of KOPRA.

High‐resolution mid‐IR limb emission spectra were recorded during a flight of the Michelson interferometer for passive atmospheric sounding, balloon‐borne version (MIPAS‐B) from Kiruna, northern Sweden (68°N) on March 14/15, 1992. These spectra are affected by the Mt. Pinatubo stratospheric aerosol, which caused an enhanced continuum emission, especially in spectra of low tangent altitudes. Aerosol extinction coefficients were retrieved from MIPAS‐B spectra at approximately 60 spectral positions in the 750–980 cm−1 and 1180–1380 cm−1 spectral ranges. Retrieved aerosol extinction coefficients range from 6×10−4 km−1 to 3×10−3 km−1 in tangent altitudes 11.3 km and 14.5 km and from 5×10−5 km−1 to 1×10−3 km−1 in 16.1 km. Their distinct spectral shape indicates the presence of H2SO4‐H2O droplets. Compositions and size distribution parameters were retrieved by least squares fitting of Mie‐generated spectral extinction coefficients to the ones derived from the spectra. Estimated spectral single‐scattering albedos between 0.08 and 0.3 indicate the significance of thermal multiple scattering. Multiple‐scattering corrections led to an increase of spectral extinction coefficients by 5–50% with highest changes at lowest tangent altitudes. Accordingly, estimated volume densities have increased by 4–20% to values of 3.66, 2.85, and 0.93 μm3 cm−3 for tangent altitudes 11.3, 14.5, and 16.1 km, respectively. Retrieved H2SO4 weights of 66–70% are in good agreement with values derived from stratospheric temperatures and water vapor partial pressures. Estimated surface densities are systematically low in comparison with in situ size distribution measurements. This finding is explained by the underestimation of small particles by the use of a monomodal size distribution in the analysis. Retrieved effective radii of up to 0.8 μm were found to be consistent with the temporal evolution of the Mt. Pinatubo aerosol.

Broad-band high resolution mid-IR limb emission spectra will be measured by the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) on the European research satellite Envisat. The presence of stratospheric aerosol is expected to cause an enhanced continuum signal in the spectra which can be used to retrieve aerosol properties. The accuracy of the retrieval of spectral extinction coefficients in dedicated microwindows is estimated on the basis of simulated radiance spectra. The feasibility of inference of microphysical aerosol parameters from retrieved spectral extinction coefficients is investigated on the basis of Mie calculations and the estimated uncertainties of spectral extinction coefficients. In this paper a quantitative error analysis of microphysical aerosol parameters is presented for different typical stratospheric aerosol loadings

High-resolution mid-IR limb emission spectra were recorded during a flight of the Michelson interferometer for passive atmospheric sounding, balloon- borne version (MIPAS-B) from Kiruna, northern Sweden (68øN) on March 14/15, 1992. These spectra are affected by the Mt. Pinatubo stratospheric aerosol, which caused an enhanced continuum emission, especially in spectra of low tangent altitudes. Aerosol extinction coefficients were retrieved from MIPAS-B spectra at approximately 60 spectral positions in the 750-980 cm - and 1180-1380 cm - spectral ranges. Retrieved aerosol extinction coefficients range from 6 x 10 -4 km - to 3x10 -s km - in tangent altitudes 11.3 km and 14.5 km and from 5x10 -s km - to lx 10 -s km - in 16.1 km. Their distinct spectral shape indicates the presence of H2SO4-H20 droplets. Compositions and size distribution parameters were retrieved by least squares fitting of Mie-generated spectral extinction coefficients to the ones derived from the spectra. Estimated spectral single-scattering albedos between 0.08 and 0.3 indicate the significance of thermal multiple scattering. Multiple- scattering corrections led to an increase of spectral extinction coefficients by 5-50% with highest changes at lowest tangent altitudes. Accordingly, estimated volume densities have increased by 4-20% to values of 3.66, 2.85, and 0.93/m s cm -3 for tangent altitudes 11.3, 14.5, and 16.1 km, respectively. Retrieved H2SO4 weights of 66-70% are in good agreement with values derived from stratospheric temperatures and water vapor partial pressures. Estimated surface densities are systematically low in comparison with in situ size distribution measurements. This finding is explained by the underestimation of small particles by the use of a monomodal size distribution in the analysis. Retrieved effective radii of up to 0.8/m were found to be consistent with the temporal evolution of the Mt. Pinatubo aerosol.

Aboard the European ENVISAT polar platform, the MIPAS (Michelson Interferometer for Passive Atmospheric Sounding) i.r. spectrometer will scan across the limb in order to record high resolution emission spectra. In the course of the definition of micro-windows for retrieval of line of sight, temperature and trace constituents, the spectral and altitudinal regions where CO2 Q-branch line mixing has to be considered have been identified. Line-by-line modelling of spectra was performed taking account of line mixing and resulting spectra were compared to those calculated within purely Lorentzian pressure broadening. The accuracy of the Rosenkranz approximation was tested and found to be sufficient in most spectral regions. The impact of CO2 Q-branch line mixing on the retrieval was compared to typical random errors due to spectral noise. Systematic errors due to the neglection of line mixing proves to play no important role for the temperature, pressure and trace constituents retrieval in spectral regions of more than 2 cm−1 distance to CO2 Q-branch centres. Apart from a few exceptions retrieval errors due to the neglection of line mixing are negligible for the spectral regions assigned for on-line processing of MIPAS measurements.

Two sets of subroutines for calculating absorption cross section spectra and transmission or radiance fields are presented. These libraries will be used in the operational data processing of the MIPAS/ENVISAT level-2 off-line processor and are part of the Karlsruhe Optimized and Precise Radiative transfer Algorithm (KOPRA). The objective in developing these libraries was to accommodate flexibility and simplicity in use without substantial loss of accuracy and efficiency.The ADDLIN library uses an efficient method for calculating absorption cross section spectra line-by-line to arbitrary high numerical accuracy. Computational efficiency is achieved by calculating each spectral line on its own optimum set of sampling points. Absorption cross section spectra are stored on flexible, non-equidistant frequency grids.The TRANSF package provides integration routines that can be quickly configured for a variety of specific applications and measurement scenarios. The routines operate on the non-equidistant frequency grids produced by ADDIN and allow the researcher to implement the radiative transfer in a simple and almost natural way. Computational efficiency results from the reduced number of sampling points on the non-equidistant frequency grids compared to equidistant frequency grids of uniform intervals.

The new Karlsruhe Optimized and Precise Radiative transfer Algorithm (KOPRA) is a line-by-line model for use in retrieval processors for atmospheric observations. It simulates infrared spectra by taking into account physical properties of the atmosphere (e.g. refraction for non-spherical earth, horizontal gradients, non-local thermodynamic equilibrium, line mixing, pressure shift) and of the instrument (instrumental line shape, finite field of view, calibration correction). Besides spectrum calculation, KOPRA has the capability to determine the derivatives of the spectrum with respect to many retrieval parameters (temperature, horizontal temperature gradient, volume mixing ratio, horizontal volume mixing ratio gradient, vibrational temperature, background continuum signal, line-of-sight, etc.). Comparisons between analytical and numerical derivatives, which are generally within a few percent, demonstrate that approximations due to run time optimized implementation are small. Furthermore, a flexible scheme is presented for handling various parametrizations of atmospheric profiles as implemented in the code in order to support different retrieval approaches.

High resolution limb emission spectra were measured by the balloon-borne FTIR spectrometer MIPAS-B in the Arctic vortex in March 1992. These spectra are significantly affected by the Mt. Pinatubo stratospheric aerosol. A method has been developed for separating the radiance signal of emission lines of trace gases from the aerosol continuum. By applying this method to the MIPAS-B spectra aerosol extinction coefficients have been retrieved at approximately 60 spectral positions in the 750 - 980 cm(-1) and 1180 - 1380 cm(-1) spectral ranges. The spectral shape of the mid-IR aerosol extinction has been derived for the tangent altitudes 11.3 km; 14.5 km and 16.1 km. On the basis of Mie-calculations it has been demonstrated that the spectral aerosol extinction coefficient is sensitive to the aerosol composition as well as to the particle size distribution. An algorithm has been developed to retrieve microphysical parameters by least-squares fitting of the derived spectral extinction coefficients to Mie-generated extinction coefficients. The retrieved aerosol parameters indicate the significance of scattering even in the mid-IR. Related corrections on the basis of multiple scattering calculations were performed. Retrieved compositions and effective volume densities of the Pinatubo aerosol in the Arctic vortex are presented.

The accurate knowledge of pressure and temperature profiles is a precondition for the retrieval of trace gas profiles from limb emission measurements. A method was investigated which allows the pressure and temperature retrieval from limb emission spectra as expected to be measured by the MIPAS-ENVISAT instrument. Regardless which retrieval scheme will be used: the simultaneous retrievability of pressure and temperature depends largely on the proper selection of microwindows. Microwindows which contain a large amount of information on these target quantities while being insensitive to systematic errors are considered to be the most appropriate ones. A microwindow selection which minimizes the pressure temperature retrieval error has been carried out for the instrument specifications of MIPAS-ENVISAT. Errors under consideration were random noise, calibration uncertainties, and neglection of possible breakdown of thermodynamic equilibrium.

MIPAS (Michelson Interferometer for Passive Atmospheric Sounding), a high resolution limb sounder, is now under construction as an ESA Developed Instrument (EDI) on ESA's planned environmental satellite ENVISAT-1. The feasibility of detecting 27 trace constituents with MIPAS in the middle atmosphere has been studied under a contract with ESA on the basis of simulated limb emission spectra in the mid-i.r. spectral region from 4.15 to 14.6 μm assuming a spectral resolution of 0.05 cm-1. Synthetic emission spectra have been generated for complete limb sequences. For the best suited spectral features, radiance profiles have been calculated for various atmospheric conditions. Based on the present performance parameters of MIPAS, altitude ranges have been estimated in which each constituent can be measured within defined limits of accuracy. The working procedure and the main results of the study on MIPAS' capability to measure the atmospheric composition are presented. While the results are specific to this instrument, they might be of general interest to investigators working in the area of high resolution atmospheric remote sensing.