Print

Progress beyond the state-of-the-art

This project will build upon results achieved in former EU projects like COSE (Compilation of atmospheric observations in support of satellite measurements over Europe’, ENV4-CT98-0750, 1998-2000) and the EU FP6 Integrated Project GEOmon (Global Earth Observation and Monitoring, 2007-2011; http://www.geomon.eu), and it will exploit the existing NDACC observational infrastructure in Europe and beyond.


Data delivery

NDACC Principal Investigators are committed to submit their data to the NDACC Data Handling Facility (DHF) twice a year, with a delay not larger than 2 years after the measurement. In the GEOmon Activity ‘Stratospheric Ozone and Climate’, we set a first step towards more rapid data delivery from NDACC stations. However, accepted delays for data submission were still of the order of 3 months after the measurements, and every partner was allowed to submit the data in a totally free format, without any metadata. The submitted data could be retrieved from an ftp server accessible via the GEOmon data portal hosted by the Norwegian Institute for Air Research (NILU).

In NORS, we will move to a data submission scheme with a delay not larger than one month after the measurement, and we will enforce a common format including metadata, in order to provide a high-quality and useful service to GAS. NORS will adopt a format that is compliant with GEOMS (Generic Earth Observation Metadata Standard, published on AVDC Web pages (http://avdc.gsfc.nasa.gov/index.php?site=1178067684, 2011) established lately in GECA, in agreement with ESA, the NASA Aura Validation Data Center (AVDC), the Envisat Validation Data Center (EVDC) hosted at NILU and the NDACC DHF, and approved by the Committee on Earth Observation Satellites (CEOS) Working Group on Calibration and Validation (WGCV). These metadata guidelines are actually implemented using the Hierarchical Data Format (HDF4 and HDF5 file formats), but are not limited to these formats; the network Common Data Form (netCDF-4) is of interest as this shares the same data model with HDF5. HDF has become the de-facto satellite data exchange format for the ESA and the NASA Earth observation missions, while netCDF is a standard in the climate research community. In NORS, we will verify whether the GEOMS-compliant HDF format satisfies our requirements. If not, we will suggest amendments to the GEOMS managers in order to optimize the implementation of the guidelines for the needs of NORS and GAS.

NORS will also deliver re-analysed time series of atmospheric measurements back to 2003 or to the start time of the observations if this is later than 2003, in support of the re-analysis performed in GAS.


Data characterization

The second important activity in NORS is the advanced characterization of the NORS data products. Work in this direction has been going on in the various NDACC Working Groups1 but a coordinated approach is required to tailor and document the products in agreement with the special needs of the GAS users. In particular, more studies are needed to assess the uncertainties associated with the NORS products, their vertical sensitivity and resolution, and their agreement with satellite data products. These characteristics must be taken seriously when comparing data from different sources, e.g., ground-based remote sensing data with satellite data or with model analyses. Therefore, in this project it is aimed at providing all information needed on vertical and horizontal averaging, location of the remotely-sensed air volume, and resolution of the data, in the form of metadata and user guides in a consistent format for all instruments. This integrated approach constitutes a significant progress beyond the state of art.

Remote-sensing data products (satellite as well as ground-based) have a multi-dimensional character, which varies with the measurement geometry, the spectral range, the species etc. In GEOmon, pioneering research was done to better characterize this multi-dimensional character and the resulting issues to be addressed when comparing two data sets with a very different perception of the atmosphere. E.g., for the first time, two-dimensional averaging kernels were derived for MIPAS profile retrievals, enabling the characterization of the MIPAS horizontal resolution and consequently a better understanding of both horizontal and vertical smoothing and sampling issues (von Clarmann et al., 2009). The latter issues have direct impact on the interpretation of the MIPAS data, on their ground-based validation and on their ingestion by data assimilation systems like the one coupled to the IFS at ECMWF. This innovative research will be extended in NORS to all the data types involved. As a result, the exact location and extent of the probed air volume, as a function of the date/time of the measurement, will be determined and included with the data products or the accompanying metadata. This will enable reducing collocation and smoothing errors in data comparisons and better understanding the representativeness of the data with respect to a model or analysis field.


Comparisons between datasets from different sources

NORS will also make progress in the evaluation of data products coming from various sources. For example, tropospheric columns from different instruments (MAXDOAS, DOAS, FTIR) were compared for the first time in GEOMon and the results showed a promising potential for satellite validation. As instrumental techniques become more advanced and retrieval algorithms more sophisticated, new data products are generated and it is important to know how they compare to older versions of the same product, or to a similar product delivered by another type of instrument or coming from another network. An example is the cross-validation between the NDACC and the more recent TCCON methane products. In the data characterization studies, we will include some NDACC products that are not mature yet for operational data delivery but that have a significant potential for the future and that are also highlighted as priority parameters in the GAS Implementation Group Final Report, like the NO2 column and formaldehyde in the mid troposphere from FTIR measurements or tropospheric ozone from MAXDOAS instruments.

Much progress was already achieved in GEOmon concerning the evaluation of the agreement between NDACC and satellite products, at various time and space scales. This work and work going on in parallel to NORS, will not be repeated but complemented with similar studies for additional data products, and for more recent versions of the satellite and/or ground-based products. Based on the various inputs regarding satellite validation, we will make a clear and complete compilation of the results in order to give the providers of GAS products a better insight into the consistency between the various datasets that are used in the generation and validation of their products. Such a compilation is not readily available at this moment.


Data integration

Having data products for the same atmospheric species but with different characteristics coming from different sources, it is possible to combine them to an integrated product carrying the best of the information from all sources. This is a very acute issue for ozone where multiple sources exist for column and profile information and where a combined product covering the O3 profile from the surface to the upper atmosphere would be very welcome. This was never achieved before and will be investigated in NORS.

Ground-based remote-sensing data are the ideal intermediate between the in situ surface concentration data and the satellite integrated column data. Research will be carried out to investigate this link, using information on data representativeness and including knowledge derived from chemistry-transport models.

Analyses of site representativeness have been carried out since several years applying Lagrangian Particle Dispersion Models (LPDM) in backward mode to establish source-receptor relationships.

Such studies were used to describe the area influencing measurements at Jungfraujoch (Folini et al., 2008) but also to inter-compare sites in the European domain (Folini et al., 2009). Different surface station categories based on parameters describing representativeness were developed within the FP6 project GEOmon (Henne et al., 2010).

Building on this experience we propose a more advanced approach within NORS comparing tropospheric column data as retrieved from ground-based and satellite remote sensing with surface in-situ measurements. The latter ones offer a solid ground for validation, since they are taken at numerous stations of extensive national networks and can be traced back to international gravimetric concentration standards. When comparing surface in-situ data with tropospheric column data, additional information on the tropospheric profile shape is necessary to link the surface concentrations to the column. The profile information is typically taken from an atmospheric chemistry-transport model with rather coarse horizontal resolution. Here we will use two independent model products. On the one hand, output from the global state-of-the-art chemistry transport model (CTM) MOZART-4 (http://www.acd.ucar.edu/gctm/mozart/) for ozone, nitrogen dioxide, and carbon monoxide, and data from the global CTM TM5 (http://www.phys.uu.nl/~tm5/), which is specialized for methane, will be used. Methane is not treated explicitly in the MOZART-4 run. On the other hand, the re-analysis products produced by GEMS and MACC (http://www.gmes-atmosphere.eu/data/) will be considered.

An additional uncertainty is introduced in the comparison if the in-situ observations are not representative in terms of model surface concentrations and satellite columns. Representativeness here is understood as a tolerable difference between the point measurement of the in-situ site and the volume average as it is represented in the model or satellite product. This mismatch is mostly driven by recent surface influences on the concentration field. By applying higher resolution backward LPDM calculations for the different sampled or simulated air masses, the mismatch between point measurement and model grid box can be quantified and either be used to filter out situations with large mismatch or to assign additional uncertainties in a quantitative comparison (e.g. weighted regression analysis). Additional representativeness issues arise for sites in complex terrain. Here the model surface height often underestimates the altitude of the actual sampling site (classically placed on mountain tops or saddles). Under such circumstances it is not a-priori certain which model grid box should be used for comparison with surface in-situ data. Assessing the surface influence through LPDM calculations for different grid boxes above the surface and comparing these to the estimated influences of the point measurement, will allow selecting the model output altitude with the best match. Detailed comparison studies including analyses of representativeness are proposed for the site Jungfraujoch that is situated in the central Swiss Alps but frequently subject to pollution episodes originating from the densely populated areas towards the South (Po Valley) and North (Swiss Plateau). For sites that are less affected by recent surface fluxes (usually island sites or sites in sparsely populated regions) a direct comparison between model-extended surface in-situ data, remotely-sensed surface concentration data and tropospheric column data might be feasible without further analysis of representativeness. This question will be evaluated for at least one more European site.

Therefore, based on existing long-term time series (ground-based remote sensing and in-situ) of established networks (e.g. NDACC or GAW) combined with model information, long-term products with better defined uncertainties (as explained above) can provide long-term information to assess satellite data on a regular basis during the whole life time of the satellite.


Operational validation of GAS products using NORS data sets

Current projects engineering the future GAS give proper importance to aspects of data quality and service quality. However, they tend to rely only on existing operational networks with fast delivery capabilities. A consequence is that they lack validation sources for several chemical species of high priority, not covered by these fast delivery networks. Several of the missing validation sources can be provided by the complementary types of instrumentation operated within NDACC. Moreover, the different types of NDACC instrumentation offer complementary perceptions of atmospheric composition, a more comprehensive view than can be offered by a network exploiting a single measurement technique. Furthermore, NDACC strives at the excellence needed to detect subtle changes in atmospheric composition and its link with climate, through a coherent series of instrument and data protocols. This makes NDACC a natural source of validation data and validation expertise.

To our knowledge, NDACC data have not been used yet for the validation of GAS products. Only a few validation tests have been carried out in the framework of the former FP6 GEMS and ESA GMES Service Element PROMOTE projects. NORS will be the first coordinated effort to provide NDACC data specifically tailored for GAS validation purposes, and to develop and implement a dedicated web-server for validating GAS products using these NORS data products on an operational basis and in a consistent way. This represents a big step forward compared to the actual state-of-the-art. The validation procedures used in NORS will be compliant with the GAS Validation Protocols being developed within MACC and PASODOBLE, which are themselves compliant with the higher level Quality Assurance Framework for Earth Observation (QA4EO) established for the GEOSS. In particular, the web-based archive we propose as part of this server, replies directly to the requirement of traceability of the quality information and its long-term sustainability. To ensure mutual consistency, avoid duplication of work, and enhance dissemination of results with upstream and downstream GAS partners, interactions will be sought with the GAS projects teams performing validation tasks and with satellite data providers in charge of validation activities (e.g. the O3M-SAF which is responsible for the production and validation of EUMETSAT atmospheric composition measurements by GOME-2 and IASI).

 

References

 

1 The NDACC organisation is composed of several working groups, according to instrument type or research theme; see http://www.ndacc.org.