8.
GOME-2 Products Validation and Monitoring
|
|
|
8.
GOME-2 Products Validation and Monitoring |
The overall objective of the GOME-2 characterisation, calibration and validation activities is to ensure that, after the commissioning phase and thereafter during the mission lifetime, the GOME-2 instrument achieves its expected performance with respect to the GOME-2 requirements specification [RD10], and that the products satisfy the EPS end-user requirements specified in [RD1]. A further objective is that the GOME-2 product accuracy will continuously improve as far as possible to satisfy the evolving state-of-the-art user requirements. The following specific objectives determine the product Cal/Val activities for GOME-2:
It can be expected that correction of GOME-2 in-orbit instrument degradation will periodically require a reprocessing of the complete data set during routine operations to ensure the consistency of the long-term data record.
GOME-2 In-Orbit Verification (GOME-2 IOV) will be under the responsibility of the European Space Agency (ESA) and will be carried out in the time period from 'launch' to 'launch plus 8 weeks'. The primary objective of GOME-2 IOV is to verify that the instrument meets its functional and performance requirements. This will be achieved by exercising specific instrument operations, first via manual commanding and then using dedicated test timelines, and by analysis of raw data from both the S and X bands using dedicated test tools. Demonstration of nominal instrument performance is a prerequisite for successful GOME-2 IOV and Commissioning Phase Hand-over Reviews. In addition, GOME-2 IOV activities are expected to provide significant input to the planning of commissioning phase and routine operations. The specific functional and performance tests carried out during IOV are fully detailed in [RD6] MetOp GOME-2 In-Orbit Verification Plan.
Verification of the correct functioning of the GOME-2 instrument requires continuous instrument monitoring activities. These activities will start during the commissioning phase, specifically during in-orbit verification of the instrument function and performance, and continue during the remainder of commissioning phase and during routine operations. Furthermore, instrument characteristics such as radiance and irradiance sensitivity will change during the GOME-2 lifetime due to in-orbit degradation of the instrument. A subset of Level 0, 1a and 1b data necessary for instrument performance monitoring and for the calculation of correction factors to account for changes in the sensitivity of the instrument, will be generated by the SPA function of the GOME-2 PPF (see [RD7] GOME-2 Level 1 Product Generation Specification and Sensor Performance Assessment for further details of specific monitoring activities). These monitoring data will be further analysed and degradation correction factors derived where appropriate. The starting point for monitoring activities using internal calibration sources are on-ground reference measurements contained within the set of Calibration Key Data files described in [RD13] and [RD18].
It is not possible to fully verify and validate all GOME-2 Level 1 products without feedback from the validation of atmospheric constituent retrievals. For further details on the validation of atmospheric constituent retrievals, in support of Level 1 validation, see Atmospheric Constituent Verification and Validation. The spectral solar irradiance data and the cloud parameters may be fully validated, but spectral radiance data can only be partially validated. There are however a number of other parameters in the Level 1 product that can be verified or used for confidence checking.
Basic checks on the scan location, swath width and ground-pixel size are carried out by comparing observed values with those expected based on the scan offset from nadir, the commanded swath width, and integration times specified in the current instrument operating timeline. In addition, the Quick Look False Colour PMD images generated by the PQE facility of the GOME-2 PPF (as described in [RD7]), in which coastlines are expected to be visible, will be compared to a validated coastline map to confirm that there are no deviations between the coastline observed in the False Colour PMD image and the coastline map.
Spectral calibration measurements are taken using the Spectral Line Source (SLS) every orbit. In order to verify the spectral calibration parameters generated using the SLS, it is planned to generate an independent estimate of the spectral calibration using the Fraunhofer absorption lines visible in GOME-2 measured spectra. The solar spectrum contains well-documented Fraunhofer absorption lines in the spectral range of GOME-2 (see [SCD24] "Allen's Astrophysical Quantities"). By cross-correlating the Fraunhofer lines in a wavelength-calibrated high resolution reference solar spectrum with those in either the GOME-2 SMR or in calibrated GOME-2 earthshine measurements, an independent estimate of the GOME-2 spectral calibration may be obtained.
Measured Stokes fractions from GOME-2 can be monitored and also validated with respect to the constraints of an extreme limiting atmosphere as described in [SCD20]. An alternative method for validating the polarisation measurements is to use a vector Radiative Transfer Model (RTM), as a reference validation measurement, which fully describes the polarisation state of the incoming radiation measured by GOME-2. The utility of this as a validation method will necessarily depend on the accuracy of the vector RTM and the input parameters for the model calculations, and also the level of validation of the model itself. In practice this method may also be considered as a model validation tool. It is preferable to validate measurements made in nadir static viewing mode for this validation exercise. In addition, the validation activity will be restricted to cloud-free scenes.
An effective cloud amount and cloud top pressure is retrieved for each GOME-2 ground pixel using the Fast Retrieval Scheme for Clouds from the Oxygen A band (FRESCO), developed by KNMI ([SCD21]). The FRESCO algorithm has been independently validated for GOME-1 data by comparison with the ISCCP cloud data set ([SCD22]). The same approach will be followed for validation of FRESCO cloud products from GOME-2. Additionally AVHRR/3, also flying on Metop, will produce cloud information at a resolution of approximately 1x1 km on the ground. The swath width of AVHRR/3 is ±1447 km which provides complete coverage of the GOME-2 swath with the exception of north and south polar scanning modes which have default offsets of +46.696º (north polar) and -46.172º (south polar) with respect to nadir (see [RD8] GOME-2 Level 1 Product Format Specification). This provides a valuable data set with which to validate the FRESCO cloud products produced using GOME-2 data. However, as GOME-2 and AVHRR measure in different spectral regions they will show different sensitivity to clouds. This will complicate the comparison of the two data sets.
The solar irradiance measurements made by GOME-2, typically obtained once per day, are combined into a Sun Mean Reference spectrum which represents an average of a number of solar spectra, subject to quality control criteria, measured during the course of a solar viewing timeline (see [RD7] for more details). The SMR can be validated by comparison with reference solar spectra, obtained from one or more satellite instruments measuring in the same spectral region. Although it is advantageous to obtain coincident measurements in space and time, solar variability when measurements are normalised to an Earth-Sun distance of 1 AU is expected to be no more than a few percent between 210 and 300 nm and less than one percent above 300 nm (see [SCD23]).
GOME-2 has a number of Earth scanning observation modes in which measurements of backscattered radiance spectra are made (see Observation Modes). The validation of earthshine radiance measurements from GOME-2 is an extremely difficult task due to the inherent variability of the observed atmosphere, clouds and ground scenes, and the dependence of the measured spectrum on the viewing geometry. These data cannot be completely validated without feedback from Level 2 product validation. Further feedback will be provided by NOAA who plan to process GOME-2 Level 1b data using the SBUV/2 ozone profile algorithm. Diagnostic information produced will provide useful feedback on the quality of the GOME-2 Level 1b data product. A number of possibilities do exist, however, for the preliminary checking and validation of GOME-2 radiance spectra and these are described below.
The UV albedo as measured by GOME is calculated as a ratio of the calibrated radiance spectrum to the SMR spectrum. In the wavelength range 240-290 nm surface contributions to the back-scattered radiance are low. The UV albedo is therefore expected to be smooth, both spectrally and temporally, and can be analysed to provide an independent estimate of the accuracy of the noise and dark current corrections. This provides a preliminary confidence check on the pre-flight characterisation of the radiance and irradiance response of the instrument and on the in-flight calibration of the radiance and irradiance spectra themselves.
Furthermore, the ozone absorption peaks in the Hartley band at a wavelength of approximately 255.3 nm. It is therefore possible to select wavelengths either side of the peak which have the same absorption and therefore can be expected to measure the same UV albedo. However, this is a region of extremely low signal levels and the utility of the method will depend on the quality of measurements at these short wavelengths in-orbit. Spatial and/or temporal averaging in order to improve signal to noise characteristics are likely to be required.
Comparison of GOME-2 backscattered radiance spectra with nadir backscattered radiance spectra measured by other satellite-based sensors can provide a preliminary validation of the GOME-2 Level 1b radiance measurements. Ideally these comparisons will be carried out using measurements made in nadir static Earth observation mode (see Observation Modes) to minimise ground scene variability and additionally differences in viewing geometry. Alternatively, nadir ground pixels in Earth scanning mode could be used for the comparison. Differences in solar zenith angle (SZA) will however remain, even for nadir viewing geometry. Additionally, only cloud-free scenes should be selected. The utility of this method may be restricted to the ultraviolet region where the sensitivity to clouds and surface albedo is reduced. Good knowledge of aerosol characteristics will still be required. A statistical comparison will be considered.
Due to the intrinsic difficulty in comparing GOME-2 radiance spectra with those measured by other instruments, measured GOME-2 spectra will also be compared to back-scattered spectra simulated using a radiative transfer model. The radiative transfer model requires as input measurements of the atmospheric state (particularly O3 and NO2 but also BrO, HCHO and SO2, and aerosol information) either from independent collocated ground-based or satellite measurements if available, or from model or climatology data. Once again the utility of this method may be restricted to the UV region.
As described in Level 1b to 2 Data Processing., the operational GOME-2 Level 2 data products will be produced by the Ozone Monitoring Satellite Application Facility (O3M SAF), part of the EPS Distributed Ground Segment. The target trace gases for GOME-2 are O3 profile, total column O3, NO2, BrO, HCHO, SO2 and OClO. Aerosol Absorbing Index (AAI), aerosol optical depth and UV maps (both clear sky and including clouds and albedo) will also be produced. See Table 4.4: Expected product accuracies for operational GOME-2 Level 2 products produced by the O3M SAF for further details on expected accuracy and product quality. The O3M SAF is fully responsible for the validation of their own products, however it should be noted that there is no commitment for validation of the minor trace gas species. For further information see [O3M1] Ozone SAF User Requirements Document, [O3M2] Ozone SAF Science Plan and [O3M3] Ozone SAF Scientific Prototyping Report. O3M SAF Level 2 product validation activities are expected to provide relevant feedback for Level 1 product validation. In particular, diagnostic output from the Level 2 retrieval algorithms will contain useful information for analysis of Level 1 product quality. Similarly, information on instrument performance and Level 1 product quality, obtained during the Level 1 verification and validation activities, will provide valuable input to the Level 2 validation activities.
In addition, a restricted processing capability for the retrieval of atmospheric constituents is required centrally at EUMETSAT in support of GOME-2 Level 1 product validation. This will comprise at a minimum an ozone profile, a total column ozone, an Aerosol Absorbing Index (AAI) and a minor trace gas retrieval capability, in particular total column NO2 and BrO. This capability will be used to provide an independent estimate, on the basis of the quality of the retrieved atmospheric constituents, of the quality of the Level 1 data products.
Validation of these products will be primarily based on comparison with other independently validated spatially and temporally collocated measurements. Sources of independent validation data sets include ground-based data, satellite-based data, and aircraft- and balloon-based data.
Ground-based measurements of products relevant to GOME-2 are available from a number of networks of well-calibrated and well-validated instruments. These networks typically provide a continuous data record of well-understood quality. Ground-based measurements will form the basis of an absolute validation of GOME-2 data. Specific collocation possibilities depend on the GOME-2 overpass time for each ground-based station location as compared to the nominal observation time at the ground station. If the available collocated ground-based ozone profile measurements are not sufficient, additional measurements, timed to coincide with the GOME-2 overpass time, may be requested as part of a coordinated calibration campaign. Alternatively, use of a chemical transport model or data assimilation model may be required to interpret measurements taken at a different time of day or SZA. Measurement techniques used in ground-based networks include the following:
The observing networks which coordinate and collect routine measurements include the WMO Global Atmosphere Watch (GAW) programme, the Network for the Detection of Stratospheric Change, and the Climate Monitoring and Diagnostics Laboratory (CMDL).
Satellite-based measurements have the advantage of global coverage which facilitates validation for all seasons and latitudes. Specific collocation possibilities depend on the detailed orbit characteristics of each satellite. In the case that the data are not closely collocated in time, access to additional analysis tools (e.g. chemical transport models, data assimilation systems, trajectory models) will be required for optimal interpretation of the data comparison. Satellite-based measurements do not provide a direct validation reference measurement as they are themselves validated with respect to ground-based measurement systems. However, on the assumption that this validation has been successfully carried out, they provide an important validation data set, particularly for the examination of seasonal and temporal variability on a global scale. Several satellites, carrying instruments which measure products of relevance to GOME-2 and from which validation data may be available, are expected to be flying concurrently with Metop. These are SCIAMACHY and MIPAS currently flying on ENVISAT, SBUV/2 flying on the NOAA series of satellites, the Ozone Monitoring Instrument (OMI) to be launched on EOS Aura, currently planned for 2004, and Ozone Mapping and Profiler Suite (OMPS) to be launched on the NPOESS Preparatory Project currently planned for launch in 2006. Furthermore, the TOMS series of instruments continues, with EP-TOMS currently in operation. The availability of this data set will however depend on the operational lifetime of the instrument.
In addition to the meteorological balloons, referred to above as ozonesondes and classed here as ground-based measurements, larger balloons which require experienced personnel and dedicated launching sites can be used to carry multi-instrument payloads. Intensive balloon campaigns have been used in the past in the context of dedicated campaigns for the validation of satellite-based data or in the framework of specific scientific programmes initiated by the European Union, e.g. the European Arctic Stratospheric Ozone Experiment (EASOE) and the Second European Stratospheric Arctic and Mid-Latitude Experiment (SESAME). Aircraft are also used to carry multi-instrument payloads which are capable of making detailed measurements of ozone and other trace gases in the upper troposphere and lower stratosphere. They have also been used in the framework of dedicated campaigns. A validation campaign involving coordinated balloon launches and aircraft flights, timed to coincide with satellite overpass times and other dedicated ground-based observations requires the involvement and coordination of many scientists and institutions. It is not proposed to plan such a campaign solely in support of GOME-2 Level 1 validation. However, should such campaign data be available, either from other related scientific campaigns or in support of the validation activities of the O3M SAF, they will be used. The requirement of the O3M SAF for a dedicated measurement campaign in support of validation of the operational GOME-2 level products is not yet decided.
Numerical Weather Prediction (NWP) systems provide an estimate of the atmospheric state based on an optimal combination of measured data, obtained from many different sources, and model forecast fields. This is typically achieved by the minimisation of an objective cost function with respect to the atmospheric state which comprises both background and observation terms. Under the assumption that the data set to be validated is not assimilated within the data assimilation system, model analysis and forecast fields provide a self-consistent independent reference which takes into account constraints both from the model and from all other assimilated observations. Therefore although analysed model fields produced from a data assimilation system do not provide an absolute validation source, they can be used for monitoring the data to be validated. This allows biases and variability between different measurement types to be easily monitored on a global scale.
Validation of GOME-2 atmospheric constituents can be achieved using any of the independent validation data sources listed above. Collocated data must be selected on the basis of coincidence criteria which will vary depending on the characteristics of the individual data sets. The collocated data will be analysed per instrument and in the case of ground-based data per instrument class and/or station. Collocated ozone profile data must be sampled at the same vertical resolution. Typically this will require that the profile with higher vertical resolution is convolved with the averaging kernel of the lower vertical resolution profile. The difference between the collocated data pairs should be calculated and compared to their expected accuracy. The results will be sorted on the basis of:
and a statistical analysis carried out. For the purposes of long-term validation and trend assessment, time series of collocated data at selected locations and in the zonal mean will also be analysed. In the event that differences in spatial or temporal collocation are greater than the defined tolerances, access to chemical transport models, data assimilation models and trajectory models will be needed to aid interpretation.
As noted above, validation of O3M SAF products is under the responsibility of the O3M SAF itself. The validation activities planned by the O3M SAF are described in detail in [O3M1], [O3M2] and [O3M3]. They comprise:
Validation results from these activities and also the diagnostic quantities produced in the generation of Level 2 products will provide valuable feedback to the Level 1 validation activities. Similarly, Level 1 verification and validation activities will provide necessary input to the O3M SAF validation activities.
An EPS/Metop Research Announcement of Opportunity (RAO) to be coordinated by EUMETSAT and ESA is planned. The primary objectives of the Announcement of Opportunity are:
Further details may be found in [RD23].
It is anticipated that scientific support to GOME-2 calibration and validation activities may be provided via this Research Announcement of Opportunity, provided relevant investigations are proposed and selected.