VALIDATION OF PRODUCTS DERIVED FROM THE ENVISAT ATMOSPHERIC CHEMISTRY INSTRUMENTS

Evert Attema, Paul Snoeij, Patrick Wursteisen, Tobias Wehr

ESA/ESTEC

Keplerlaan 1, Postbus 299

2200 AG Noordwijk, The Netherlands

eattema@estec.esa.nl, psnoeij@estec.esa.nl, pwurstei@estec.esa.nl, twehr@estec.esa.nl

Rob Koopman, Claus Zehner

ESA/ESRIN

Via Galileo Galilei,

I-00044 Frascati, Italy

rkoopman@esrin.esa.it, czehner@esrin.esa.it

INTRODUCTION

Following internationally agreed conventions (by the Working Group on Calibration and Validation of the Committee on Earth Observation Satellites), `Calibration' is defined as the process of quantitatively defining the system response to known, controlled signal inputs and `Validation' is defined as the process of assessing, via independent means, the quality of the data products derived from the system outputs. Fig. 1 shows the concept: Envisat observes a scene and its raw data are subject to calibration in the ground segment.

Fig. 1. Calibration/Validation Concept

The results of the complete processing chain are estimates of geophysical variables (the so-called level-2 data products). The instruments as well as the processing algorithms must be carefully calibrated and verified during the commissioning phase. This process has to be repeated during the operational phase of Envisat as required. Validation is basically a comparison of Envisat level-2 data products and estimates of the different geophysical variables obtained by independent means, the validation instruments. Validation is closely linked to calibration because inconsistencies discovered in the comparison of Envisat Level 2 data products to well-known external instruments can have many different sources, including inaccuracies of the Envisat instrument calibration and the data calibration algorithms. Therefore initial validation of the geophysical variables provides feedback to calibration, de-bugging and algorithm improvement. If the differences observed during validation are within satisfactory limits the process of validation is completed and a final quality statement can be made.

Full validation of all data products available from the Atmospheric Chemistry Instruments on Envisat (MIPAS, GOMOS and SCIAMACHY) is quite a challenge and therefore it has been decided to adopt a step-wise approach. As a first step the intention is to arrive at a first quality assessment of the data products for near-real time distribution which include the geophysical variables shown in Table 1. This core validation is planned during the six months commissioning phase of Envisat. The results of this exercise will be presented at a validation workshop nine months after the launch. It is anticipated that more work needs to be done after this workshop on all Envisat data products both for near-real time and for off-line distribution.

Table 1. Geophysical variables included in the Level-2 data products for near-real time distribution

The algorithms designed to derive estimates of the above parameters need to be verified. For this a large number of correlative observations under a wide range of conditions is needed to arrive at a representative and statistically relevant data quality assessment, and to provide insight into sources of error both in the Envisat data and the correlative data sets (Fig. 2).

In order to achieve this within the tight time schedule the best use must be made of the available resources. For the Atmospheric Chemistry Instruments on Envisat it has therefore been decided to plan a joint geophysical validation programme that is not instrument specific but serves all three instruments. For the co-ordination of the activities the Atmospheric Chemistry Validation Team was formed (ACVT). Whereas ESA and institutes/companies under ESA contract are participants of the ACVT the main body of the work relies on Principal Investigators selected on the basis of the responses received by ESA to the Announcement of Opportunity for Envisat Science and Applications in the category calibration/validation.

Fig. 2.: Number of atmospheric chemistry validation instruments as a function of instrument type and of the

geophysical parameter they measure

The ACVT methods can roughly be categorised into different approaches and consistent with these the group is divided into different subgroups on

• balloon and aircraft campaigns
• model assimilation

• satellite intercomparison

• ground-based measurements

In addition three instrument-specific groups are associated with the ACVT dealing with algorithm verification.

The data coming from the various validation campaigns will be stored within a central data storage facility established at the Norwegian Institute for Air Research (NILU) in Norway. NILU will provide access to correlative measurements from sensors on-board satellites, aircraft, balloons and ships, as well as from ground-based instruments and numerical models, such as that of the ECMWF. Particular emphasis will be put on the quality control of such data. Users will be able to

connect with the database to add or retrieve data according to their requirements. Access to such a range of data will strengthen the statistical significance of the results and increase the chances of detecting errors in the processing algorithms. Two types of data will be stored in the NILU database, fixed point and transect data. Transect data will only be provided for inclusion in the database for selected times which correspond to the satellite overpass. Envisat data will not be stored in the NILU database although other correlative satellite data will be included to facilitate their comparison with data acquired by Envisat.

In order to test validation procedures, a rehearsal will take place in the autumn of 2000. Emphasis will be placed on data transfer procedures and the use of the NILU facility. In addition, the rehearsal will provide Principal Investigators the opportunity to test software tools and to familiarise themselves with (simulated) Envisat data products.

In subsequent sections the plans of the ACVT subgroups will be presented in some detail.

Fig. 3. Launch of a balloon from Kiruna

BALLOON AND AIRCRAFT CAMPAIGNS

The objective of the Envisat Stratospheric Aircraft and Balloon Campaign (ESABC) is to contribute to the validation of the MIPAS, GOMOS and SCHIAMCHY Level-1b and Level-2 data products. This will be achieved by using several sensors installed on aircraft and stratospheric balloons. The leading scientists in the field of stratospheric flights will participate in the campaigns.

The components of the ESABC campaign have been selected to perform as complete a validation as possible for the three Envisat Atmospheric Chemistry instruments whilst optimising the use of existing facilities and launch sites. Consequently, several sites have been selected and these are located at mid-latitudes, at northern latitudes and near the Equator. The flights have been organised to ensure the measurement of atmospheric constituents during several seasons. The acquisition of data at northern latitudes in the Arctic vortex during the winter is of great importance and has been given special attention. This strategy also implies that the ESABC activities span beyond the validation activities performed during the commissioning phase.

The flight programme comprises large and small balloons as well as high altitude aircraft, the German Falcon and the Russian M-55, also called Geophysika. The first campaign will take place in Aire sur l'Adour, France, and will comprise the launch of two SAOZ payloads. Such payloads can be launched by relatively small balloons of 15 000 m ³. A similar campaign will take place in Nov-Dec 2001 in Bauru, Brazil.

A large balloon campaign will take place in Kiruna (Fig. 3), Sweden, from January to March 2002. The key objective of this campaign is to acquire data from the Arctic vortex and to compare the campaign data to Envisat data, thus validating the algorithms in the high-latitude regions. Another large flight campaign will take place in Aire sur l'Adour during September and October 2002. Balloons of 150 000 to 400 000 m ³ will be used during the two large campaigns. The payload of the large balloons will be composed of a number of different instruments both in-situ sampling instruments and remote sensing instruments.

In addition to these two major balloon campaigns additional balloon flights will be performed from the site of Trapani, on the island of Sicily, Italy.

Compared to stratospheric balloons, which can reach higher altitudes, aircraft have the ability to fly for many hours at an altitude of approximately 20 km and are able to perform flights spanning several thousand kilometres. In addition, aircraft possess high flexibility to achieve close temporal and spatial coincidence with satellite overpasses almost anywhere on the globe and under most weather conditions. As in the case of the large balloon campaigns, two aircraft campaigns are planned, one during the winter and the other in late summer 2002.

Two carriers will participate in the ESABC. The first aircraft is the meteorological research aircraft Falcon 20 operated by the German Aerospace Centre (DLR). It is a well-established research platform that can carry a payload of 1100 kg at a maximum altitude of 13 000 metres. The payload is composed of a radiometer, an Ozone Lidar and a spectrometer. For each campaign, the aircraft will perform flights from its home base in Munich, Germany, to Kiruna and Greenland or to the Seychelles.

The second contributing aircraft will be the high-altitude plane M-55 depicted (Fig. 4). The M-55 can carry a payload of 1500 kg to an altitude of 22 km. Its endurance of over 6 hours makes it possible to perform long flights over large areas. The payload of the M-55 is composed of a dozen sensors well adapted to the validation of GOMOS, MIPAS and SCIAMACHY. Two campaigns are planned during February and March 2002 from Forli, Italy, to Kiruna, and a mid- latitude campaign in the summer 2002 at Forli.

Fig. 4.

The Russian high-altitude M-55 aircraft

MODEL ASSIMILATION

The aim of data assimilation techniques is the combination of theoretical models and sparse measurements for the forecast or analysis of the state of the atmosphere. In particular, numerical weather forecast assimilation techniques are used to combine measurements from satellites, balloons, ground stations etc., with weather models in order to predict and analyse the state of the atmosphere at a given position and time. Today, in atmospheric chemistry research, data assimilation techniques are also being developed which will enable the exploitation of information that would not be available from models or observations alone.

The assimilation efforts will be organised into two main activities:

·

Assimilation into Numerical Weather Prediction (NWP) models. These will be performed by operational meteo entities such as the European Centre for Medium-range Weather Forecasting (ECMWF). These types of models assume a neutral atmosphere (i.e. no chemistry details are involved) and use a proven operational assimilation technique.

·

Assimilation into Chemical Transport Models (CTM). These are more experimental in nature but include a representation of the atmospheric chemistry.

In addition, as a service to other ACVT participants, extracts of the model assimilation analysis results for a requested location and time are provided. This service could be used by operators of ground-based instruments to compare their results with Envisat measurements interpolated to the specific location and time of the ground-based measurements.

Assimilation into Numerical Weather Prediction Models

Envisat products relevant to meteorological applications will be analysed using the ECMWF data assimilation system. Detailed statistical analyses will be made of the differences between the Envisat products and ECMWF's assimilation field for the corresponding geophysical quantities. This will assist the Envisat instrument teams in the characterisation of errors and biases of the data products and will also help ECMWF to characterise errors and biases in the model. In addition, ECMWF will provide, for ESA, a long-loop monitoring capability to help maintain the quality and integrity of the Envisat products.

Assimilation into Chemical Transport Models

A sequential assimilation system is proposed based on one originally developed for GOMOS ozone and which has recently been extended to assimilate MIPAS and SCIAMACHY ozone products. The data will be assimilated in a global chemistry- transport model (CTM). The chemistry considers a comprehensive set of species and reactions. The necessary dynamic inputs will come from ECMWF. The following activities are planned:

·

Assimilation of GOMOS, MIPAS and SCIAMACHY ozone products independently in order to perform an inter- comparison between the resulting analysis fields. This will allow the comparison of ozone measurements acquired by the three instruments regardless of their different observation geometry's and viewing directions.

·

Geophysical validation with independent sources. GOMOS, MIPAS and SCIAMACHY ozone products will be assimilated and compared with independent measurements from the ground (e.g. ozone measurements from NDSC stations) and from satellites (e.g. GOME). The assimilation will be used to interpolate the respective Envisat measurements in space and time to the location of the validation measurements.

A different system proposed uses a 4D-VAR assimilation approach. The 4D-VAR strategy is fundamentally different from a three-dimensional sequential assimilation, since it takes into account the temporal evolution of the state of the atmosphere within a given analysis time window. In a sequential scheme, a newly analysed field value is computed by an optimal interpolation between the model and the measurement, weighted by their respective errors. The analysis field is then propagated forward in time until the next measurement, when the procedure is repeated. In the 4D-VAR approach, a time interval is chosen. A first-guess model trajectory is computed by running the model from time 0 to time T. During this run, the differences between model calculation and measurements are recorded. A second model run is performed with the `adjoint model', which transports the mismatches backwards in time in order to estimate an improved start value for a new model run within the time window. This iterative procedure is repeated until convergence is reached. The final results will therefore not show jumps within the time window and optimum consistency with both the model and data will be assured.

The 4D-VAR approach will particularly help to constrain species in the analysis according to their chemistry model relation and interaction with other species. This system can therefore be helpful in the examination of the consistency between measurements of different species, e.g. GOMOS measurements of ozone and NO₂. Time series and / or total columns from the assimilation analyses of the measured species will be provided.

The main goal of the technique is operational assimilation of Envisat atmospheric composition measurements. This consists in independent assimilation of species profiles for GOMOS, MIPAS and SCIAMACHY separately. Global maps of the measurements will be created and biases analysed. Rather than near-real time data, quality-checked data will be used in order to ensure the highest quality.

SCIAMACHY ozone measurements will also be validated through assimilation into a system presently being used for GOME ozone assimilation. The GOME assimilation model uses the 4D-VAR technique and a global tracer advection model. In the lower stratosphere, most ozone change on time scales of one or a few days is due to transport. Therefore, ozone can be used as a tracer of air masses. The model simplifies the ozone transport problem assuming that a single two-dimensional latitude-longitude wind field can describe the evolution of ozone columns. The assimilation therefore uses the wind field at the altitude where most of the ozone variability occurs, typically around 10 - 20 km. High quality wind fields are taken from the ECMWF model. The tracer transport model is run at a high resolution of 100 x 100 m ², a resolution comparable to the size of the GOME ground pixel.

This assimilation application focuses on the atmospheric dynamics rather than on the atmospheric chemistry. The system is being operationally used for GOME ozone column assimilation. Fig. 5 shows an example of the assimilation of GOME ozone columns in the Southern Hemisphere under ozone hole conditions.

Fig. 5.

The ozone hole of September 11, 1996. The southern hemisphere GOME total ozone data used as input for the assimilation are shown in the left plot. The middle plot shows the assimilated field at 12 GMT. The corresponding uncertainty distribution is shown in the right-hand plot. All scales are in Dobson units. (Courtesy KNMI, De Bilt, The Netherlands)

The model will be adapted and further developed to assimilate SCIAMACHY ozone profiles. Near real time assimilation is planned using ECMWF forecast wind fields. For SCIAMACHY validation, two- and three- dimensional ozone fields including their errors will be provided.

SATELLITE INTERCOMPARISON

The comparison of equivalent products from other satellite systems is an effective method of validating products on a global scale. A variety of products will be validated including, at Level-1, irradiances, radiances, reflectances, and polarisation measurements, and at Level-2, trace-gas columns and profiles, aerosols and cloud detection. Data from AMSR/ADEOS-II, AMSU/NOAA, AVHRR/NOAA, ATSR2/ERS2, GOME/ERS2, HALOE/UARS, HIRLDS/EOS, IUE, METEOSAT, MLS/UARS, MOPITT/EOS, OMI/EOS, OSIRIS/ODIN, POAMIII, POLDER, SABER, SAGEIII/EOS, SBUV, SMR/ODIN, SOLSPEC/Alpha, and TOMS will be used.

The intercomparison of large samples of coincident satellite measurements will enable a global validation under different observational and atmospheric conditions. This will result in the estimation of product and retrieval algorithm accuracy. Analyses of time series will be used to detect potential time varying biases between different instruments. This activity will furthermore contribute to the establishment of consistent long-term global data sets (e.g. ozone), which combine observations from different satellite instruments.

Assimilations of TOMS and SBUV/2 ozone data will be used for comparison with GOMOS, MIPAS and SCIAMACHY ozone Level-2 products, rather than for assimilating Envisat measurements. This approach is highly complementary to the assimilation approaches described previously, because the original GOMOS, MIPAS and SCIAMACHY Level-2 data products themselves are the basis for comparison. The relative time and location of ozone fields from the assimilation of TOMS and SBUV/2 will be used with the respective GOMOS, MIPAS and SCIAMACHY ozone measurements for direct comparison.

GROUND-BASED MEASUREMENTS

Networks of ground-based instruments and sonde launch sites will provide a suite of correlative measurements covering a wide range of geophysical conditions. The aims are to generate a large number of data sets for intercomparison with GOMOS, MIPAS and SCIAMACHY Level-2 products. Most ground-based spectrometers are operated routinely, and

soundings are performed between once and three times per week. Lidar instruments will be operated during "visibility" by the atmospheric sensors.

Instruments at many sites covering the globe will perform these frequent observations. Since the ground-based group will generate the largest number of coincident data sets, statistical analyses will be possible already early in the commissioning phase. The data acquired will also be used for detailed analysis of the differences found during intercomparison, and some of the parameters will be relevant for validation of Level-1 products.

Dedicated Lidar measurements will be performed as close as possible in time and space to Envisat instrument observations. For routine measurements, collocations will be identified a posteriori from the Envisat and correlative data sets. Trace gas columns from ground-based instruments will be compared to columns and integrated profiles from Envisat. For ground- based profiles and soundings, further pre-processing is needed to take into account the differences in altitude resolution. Both a qualitative assessment of GOMOS, MIPAS and SCIAMACHY accuracy and precision, and an investigation of deviations will be performed.

Sonde Measurements

Ozone and PTU sondes will provide routine sets of data for use in Envisat validation activities. Most of the validation opportunities relate to Level-2 products. Whilst ozone sondes are subject to routine launch by meteorological offices, there are opportunities for additional launches to meet the validation requirements during the commissioning phase. Ozone sondes are insensitive to the presence of clouds.

Spectrometers

The Network for the Detection of Stratospheric Change (NDSC) and other participants will provide more than seventy- seven UV-VIS spectrometers. These spectrometers will generate measurements of total O ₃ , NO ₂ and, in unpolluted areas, the integrated stratospheric column of NO ₂. Amongst these, nineteen instruments of the SAOZ-type will provide preliminary data in near real time. Dedicated UV-VIS spectrometers will measure the column abundance of BrO and OClO. Total O ₃ will also be measured through the use of twenty-two Dobson and Brewer spectrophotometers (Fig. 6).

Nine DOAS instruments will provide BrO and OClO columns in addition to O ₃ and NO ₂.

Radiometers

Twenty-one ground-based microwave radiometer instruments (MWR) will contribute to the Envisat validation activities. These are capable of retrieving O ₃ , H ₂ O, ClO, and most of them also retrieve temperature. They are highly stable and O ₃ retrieval is largely unaffected by clouds. The disadvantages are that measurements can only be made for altitudes above 12 km and that data integration times for some species can take several hours. Ozone and temperature can be retrieved from about 12 km up to 55 km and for some instruments even up to 80 km.

Lidars

The NDSC and other participants will provide routine measurements of ozone, temperature, aerosol backscatter and extinction, and cloud parameters, as a function of altitude. Twenty-two ozone lidars will provide profiles of O ₃ and temperature, among them Raman systems that also retrieve aerosol profiles. Four additional Lidars will be dedicated to the retrieval of aerosol parameters and two systems have been constructed to retrieve water vapour. Lidars can be tuned to match specific validation requirements and they cover a wide range of altitudes; O ₃ can be retrieved up to 50 km altitude, dedicated systems are capable of retrieving temperature up to 90 km, and aerosol and cloud backscatter can be acquired up to 95 km at resolutions varying from 150 to 300 m. Stratospheric ozone lidars are only usually operated at night, but some lidars have been adapted for daytime use. In daytime mode these systems have a lower sensitivity due to the higher straylight intensity.

Fig. 6.

Brewer spectrophotometer

Fourier Transform (FTIR/FTS) Spectrometers

Three UV-VIS-IR Fourier-Transform absorption Spectrometers (FTS) and nineteen Fourier-Transform Infrared Spectrometers (FTIR) will be deployed. The FTS instruments will provide not only columns of O ₃ , H ₂ O, NO ₂ , N ₂ O, CH ₄ , and CO , but also profiles of these trace gases. The FTIR instruments are capable of retrieving columns of H ₂ O, H ₂

CO, CH ₄ , N ₂ O, O ₃, HNO ₃ , ClO, NO₂ , and additional species that will indirectly support the validation activities. In addition, vertical profiles of O₃, CH ₄, HNO₃ and N₂O will be experimentally retrieved.

Other Instrumentation

Sun photometers and GPS receivers will be used to validate products such as water vapour and aerosol column data.

CONCLUDING REMARKS

In this paper the planned geophysical validation of the data products from the Atmospheric Chemistry Instruments on Envisat (MIPAS, GOMOS and SCIAMACHY) were presented. The information is based on the proposed and ongoing activities of the members of the ACVT subgroups. The authors of this paper wish to acknowledge the contributions of their team members to the paper and to the Envisat validation.

Keywords: ESA European Space Agency - Agence spatiale europeenne, observation de la terre, earth observation, satellite remote sensing, teledetection, geophysique, altimetrie, radar, chimique atmospherique, geophysics, altimetry, radar, atmospheric chemistry