SUMMARY OF SESSIONS

Summary of thematic sessions

• Soil Moisture
• Hydrology
• Land Use
• Forestry
• Digital Elevation Models (DEMs)
• Geology
• Hazards
• GOME
• Atmosphere using ATSR, PRARE and GOME data in synergy
• Ice Properties and Ice Sheet Topography
• Sea Ice
• Ice Dynamics
• Coastal Zones, Ocean Dynamics and Features
• Global Change
• Winds
• Waves
• Sea Surface Temperature
• Ocean Circulation
• Altimetry
• Marine Geoid
• Meteorology
• ATSR Instrument Performance
• Orbits
• ENVISAT Special Session

Summary of working sessions

• Atmospheric Artefacts in SAR interferometry
• Observation of Landcover from ATSR data
• JERS/ERS Synergy

Soil Moisture

A sound theoretical basis has been established for the use of radar backscattering data from C-band SAR and scatterometers for the estimation of soil moisture. A soil moisture retrieval strategy, based on a log-linear relationship, has been developed for the use of ERS data taking into account the limitations introduced by surface canopy and surface roughness effects.

Operational applications were demonstrated over desert areas with stable roughness and over areas with sparse vegetation. Regional correction algorithms for roughness and vegetation effects were developed.

Interferometric coherence is used to delineate dielectric (soil moisture & freezing) and roughness effects. This technique requires short revisit times such as during the three-day repeat orbit and the ERS-1/ERS-2 tandem mission.

New information was presented on the monitoring of thawing/freezing phenomena by large scale observations with high temporal resolution using the scatterometer.

Archived tandem data shall be further exploited and further tandem acquisitions should be considered.

Hydrology

The entire session was devoted to the use of ERS SAR data.

SAR was demonstrated as an established tool for flood monitoring. Flooding causes sufficient image contrast for reliable detection and mapping of flooded areas without applying complex retrieval algorithms. By fortunate coincidence of the ERS overpass and the flooding event it was shown that SAR actually can see the flood wave in a river.

A reliable relationship between radar backscattering and accumulated rainfall in a region was established. ERS SAR data are being assimilated in hydrological models through soil moisture estimation derived from radar backscattering.

ERS SAR images clearly show human impact (road construction) in environmentally sensitive areas and SAR is used as an operational tool to design counter measures against these impacts. Planned future studies on the use of phase information (interferometry) and on rain effects need ERS tandem data. Limitations in this application are due to the long acquisition repeat cycle i.e. 35 days.

Land Use

Within the MARS project framework, former studies have already demonstrated the capability of ERS SAR data to drive early crop estimates. New methodologies aiming at the improvement of the system have been presented.

• A colour coded classified product is generated by the combination of 4 optimal acquisitions of ERS data during the winter and the spring seasons. Eight distinct classes of crops are discriminated with 80% of accuracy for winter combinations.
• The integration of ancillary data such as agricultural practices, meteorological data and soil types into a structured and coherent database has been tested. It allows the analysis of crop effect on the backscatter, the planning of satellite acquisitions and queries at a site and point level.
• A crop information system has been successfully developed and used over MARS test sites in UK, integrating ancillary data and satellite data for agriculture monitoring and production of agricultural statistics. This automatic system can be further customised for specific needs, as has been done for the Ministry of Agriculture in Paraguay.

The ERS capabilities to monitor specific crops is still investigated:

• The row direction of the maize cultivation was modelled and correlated to the radar backscatter. The discrimination between physiological stages was shown and extend to other crops.
• The production of composite ERS SAR images from change index maps allowed the discrimination of different rice cultivated systems but the validation of results still needs to be done.
• The mapping of agricultural and non agricultural areas from ERS SAR data has been validated in Senegal, at a regional scale. It was the first spatial presentation of agricultural acreage. Because of the low density of the vegetation and the small size of the fields, it was impossible to discriminate cotton cultivation. Further investigation need to be performed over an intensive cotton production area.

Good correlation between the backscatter and soil moisture for different crops was found when soil moisture is estimated through the Water Deficit Index calculated from NOAA AVHRR data. Further development will be performed using ATSR data.

ERS scatterometer data can also be used to estimate vegetation parameters such as the biomass. A modelling approach has been developed over the Sahel. Extension and validation to other sites in the region will be done as well as the combination of ERS scatterometer data with data from other sensors.

ATSR-2 data have been tested for mapping and monitoring of land cover. The correlation of the vegetation index and the percentage of vegetation is poor, due to the presence of a sparse vegetation. Good correlation was found through a simulation of the bi-dimensional reflectance distribution function of ATSR-2 pixels over dry land surfaces. Further methodological work is still necessary to improve the results.

The second session on Land Use was comprised of 19 presentations and 24 posters. New results were reported on post-processing in applying principal component analysis on multitemporal data sets. The 1st component represents a well de-speckled image.

The effect of the incidence angle was discussed: up to 3db was found as difference between near and far range with respect to corn fields backscatter. In wetland monitoring, the dampening effects of the vegetation was used to improve the classification. For biomass estimations with SAR the importance of comprehensive auxiliary data was stressed. For banana plantation classification SAR was reported to be perfectly adapted with 95% classification accuracy.

A interesting result was reported with ATSR-2: there is a good correlation between these data and geophysical vegetation parameters (temperature, coverage) and also grass-fires could be well detected.

Forestry

As a result of the availability of data from the 3-day repeat cycle of ERS-1 in 1994 and from the ERS-1 and ERS-2 Tandem Mission, several projects/studies in the domain of forestry have made use of Interferometric SAR information with promising results.

The recognised limitations of the radar backscattering intensity (in the ERS configuration in particular) for thematic classification have been overcome both for boreal forest and tropical forest through the use of coherence images.

At the same time, the operational perspectives of SAR interferometry techniques for forest inventory issues have been demonstrated in countries like Finland, with a long, consolidated tradition in the utilisation of optical data as a primary source of satellite information.

The understanding of the interferometric signature has improved, leading to a much wider set of applications in the domain of forest-related issues.

In contexts like boreal forest fires, the contribution of SAR can be seen rather than in the monitoring phase where AVHRR/ATSR play a major role, in the post-phenomena phase (burned areas assessment), i.e. when the fire signatures observed on SAR imagery are analysed and processed.

For what relates to the complete double coverage over Central Africa made available through the CAMP Project, further interesting developments are in progress, both for thematic applications, processing techniques and extensions to other sensors such as TM or NASDA J-ERS to pull together a network of interested partners for the CAMP Thematic interpretation.

This will ensure on short term a rapid expansion of the original, innovative concept of CAMP, i.e. to serve as a catalogue data base for further navigation and information extraction.

As to particular aspects of forestry, the Session has offered the opportunity to consider the application of microwave remote sensing to forest vitality monitoring. Here investigations are based on the information that microwave signatures can provide on one of the main tree components, i.e. water content.

Scientists and Project Leaders need to spend further effort on transferring their most significant findings into easily understandable and applicable concepts for use in an operational context.

Digital Elevation Models (DEMs)

A global DEM has been produced using the ERS Radar altimeter WAP (Waveform Altimeter Product) data from both of the 168 day repeat Geodetic Mission phases. This DEM has been interpolated to a 10km spacing grid and appears accurate on the metric scale for flat, homogeneous regions with the height error increasing rapidly in areas of rough terrain (local slopes >1 deg). Nevertheless, this DEM represents the most accurate source of topographic information today for many areas of the world (e.g. parts of S-America and the Polar Ice Sheets). A full validation of the accuracy of this global DEM is in progress.

The detailed characteristics of the RA waveform returns indicate the possibility to distinguish specific types of waveform that are related to the type of land surface involved.

Regarding the generation of high (spatial) resolution DEMs from SAR interferometry, progress has been made in the development of an automatic technique to generate the best consensus DEM from a number of ERS Tandem acquisitions, typically 5/6 pairs. This technique removes DEM outliners due to variations in the atmospheric conditions between acquisitions. The technique generates a DEM error map, with expected rms errors in the range of about 5m. The method has been applied to the test areas of Vesuvius and Etna in Italy, and first validations of the consensus DEM accuracy lead to figures of under 10 m rms, possibly due to errors in the reference DEMs used to perform the validation. Maps of the topographic error due to the atmosphere can be generated by differencing the consensus DEM with a single-pair DEM.

The fusion of DEMs from ascending and descending passes of the same area shows a clear improvement in the combined DEM quality by filling in layover and shadow regions and reducing height error by half (from 20 m to 10 m rms) even outside these regions. This analysis was performed on 3 day and 35 day repeat data for Catalunya in Spain. A corresponding analysis of Tandem data for the same region is in progress.

Several applications related uses of ERS DEMs generated by SAR interferometry were presented, including:

• Ice-mass balance determination in E. Greenland
• Climatic change history through the tilting and orientation of alluvial fans
• Coastal uplift & inland subsidence associated with tectonic movements in Chile,
• Volcanic subsidence through 3-year ERS differential INSAR in California.

Geology

6 papers and 4 posters were presented. Traditional geological mapping techniques on SAR imagery have been further developed, emphasising especially the information enhancement using ERS-SAR stereopairs, through image-summation to improve significantly the spatial resolution, as well as by applying optical/microwave merges. It can be considered now as a matured and well established application.

SAR data integration into a GIS environment has also be successfully achieved, whereas SAR is considered to contribute more to local mapping in contrast to LANDSAT TM used for regional geological evaluations.

Successful applications of SAR interferometry were reported for Siberian Pipeline Track monitoring and land subsidence measurements. The last one has reached commercial exploitation in the US for oil-well environment monitoring, despite the fact that the 35 day repeat cycle is too large for some of the earth movements to be observed.

Hazards

17 papers and 15 posters were presented. In 20 papers out of 32, the use of interferometric techniques were reported. The subject most frequently treated was tectonics (8) followed by volcanoes (6) and floods (4).

In using ERS-SAR imagery, accurate mapping for geological applications and for floods was reported.

The volcanic eruption in Iceland has turned out to be an attractive testbed for fast response from ERS SAR interferometry. To meet the user community’s requirements ESA decided at short notice to operate both satellites again for a short period of time. But also Etna remains in the centre of the attention: after having demonstrated that the whole mountain was subsiding after the most recent eruption, now also local deformations due to the weight of recent lava-flow has been observed.

With respect to tectonic movements, the creeping along the San Andreas fault could be astonishingly well mapped and quantified, and the dangerously locked part of the fault identified. Observations are also ongoing both in the Chilean Desert, comparing GPS and SAR interferometry successfully; and also in an earthquake prone zone of Southern Italy.

For land subsidence measurements e.g. in coal mining areas, using again differential interferometry the data applications are close to operational. The problems caused by atmospheric effects could largely be overcome by using multiple data sets.

For flood monitoring, the introduction of GIS is proposed in order to get information also in urbanised areas and in forests. For the mapping of flooded areas in Siberia the coherence between two acquisitions were used to improve the image-content of multi-temporal data sets.

ATSR has also been identified and proposed to monitor globally volcanic activities by using the 1.6 mm band on night-passes. Both the building-up of the temperature prior to the eruption and also post-vents such as the characteristics of the cooling of lava could be detected on volcanoes in different parts of the world.

The second part of the session on Hazards included, among other, issues covering ground deformations caused by earthquakes or volcano activity, use of ERS-1/ERS-2 SAR interferometry and differential interferometry products in studies in seismic areas, joint ERS SAR and SIR-C use for seismic risk analysis.

It appeared that most of the achievements were obtained through the experimental, combined use of all possible sources like GPS, DEMs, SIR-C, ERS INSAR and Tandem data.

New fields of investigation in the domain of hazards include measurements of elevation changes deriving from mining activities with the contribution of SAR interferometry.

The economic interest in minimising the damage at surface is likely to push these types of application towards further development.

Along the way to pre-operational demonstration projects in flooding it has to be remarked that some limitations in the use of SAR still exist. These are represented by the fact that in some instances only a static detection at a given date is possible if the flood peak is missed due to the data acquisition constraints.

The Tandem coherence images may give the additional information necessary to obtain sufficient classification accuracy as demonstrated in the case of Beziers floods in Southern France.

GOME

Instrument performance, calibration, processing

In general, the instrument is showing an excellent performance and stability. The diffuser, essential element in the radiometric calibration path, is showing within the signal-to-noise limit of the method no sign of degradation. The change in radiometric response detected early in the programme and believed to be due to coating off-gassing could in principle be confirmed to the GOBELIN measurements with the GOME Breadboard Model. Increase of dark current and noise due to accumulated radiation is within the range of expectation and does not have a significant impact on the quality of the measurements. Internal straylight levels are low and therefore hard to quantify. Further improvements are necessary for the wavelength calibration, in particular in those regions of the spectrum in which the calibration lamp provides only few, weak or uncomfortably distributed lines. Use of Fraunhofer lines to improve the wavelength registration has been demonstrated to work in principle.

For most of the identified problems, solutions have been implemented in the GOME data processor, which otherwise works routinely and stable.

GOME Level 1 Product:

GOME sun spectra are in good agreement with UARS/SOLSTICE measurements.

Better than 10% in Channel 1

better than 4% in Channels 2-4

GOME measurements have a very low noise which allows MgII index calculation to monitor solar variability.

GOME Level 2 Product (Other than O₃ and NO₂)

Non Official Products

GOME demonstrated the capability to measure trace gasses in the troposphere beside O₃ and NO₂:

SO₂ was measured in the troposphere during a volcanic eruption in Africa

B_rO can be measured on global scale in the troposphere and stratosphere.

OCLO can be measured in the stratosphere during pre-ozone hole conditions

HCHO was measured during biomass burning events over Africa and over the cities Los Angeles and Mainz. HCHO is an important species in the oxidation process in the troposphere.

Aerosols could be retrieved mainly over sea with some preliminary data over land (especially desert).

Surface reflectance: albedo in the visible can be retrieved to be used for climate models (ground pixel size of 40x80 km).

The Polarisation Measurement Devices are useful to get information about cloud cover which is important for the retrieval and studies of trace gases.

Good spectroscopy data (measured in laboratories) are needed to improve the accuracy of the retrieval of trace gasses.

UV and visible radiance/irradiance of Earth/Sun/Moon

The high precision of the solar irradiance measurements enables the derivation of MgII indices for the characterisation of solar activity. The observed long term drift in the irradiance measurements over several months might only in part be a true instrument degradation (at least for the UV regions, but this was expected) but might be caused by other effects. Work to pin down the real cause is in progress, but has to rely on further long term observations. The use of moon calibration data, limited by the moon visibility and the complicated evaluation process, may be of help on the long term to distinguish possible causes.

Total Ozone Column

Comparisons of GOME derived total ozone columns with ground based measurements performed by various networks (Dobson, Brewer, SAOZ, NDSC) show good agreement at mid-latitudes but various biases at other regions and solar zenith angles. This is confirmed by processing GOME data with an adapted TOMS algorithm. The main reasons for that have been identified to be the climatology used to calculate the air mass factors, and a wrong slit function for the convolution of the oxygen reference spectrum, leading to systematically too high fractional cloud measurements and air mass factor computations. Work to compute new air mass factor tables to overcome this problem is I progress.

The precision of the derived ozone values is excellent: from the assimilation of total ozone values into global advection models, and comparison of the advected with the true distribution, a precision of better than 2% could be quantified.

Ozone Profiles

Several groups could demonstrate the potential to retrieve ozone profiles from the GOME data. The novelty is that, because of the extended wavelength range, the range could be extended into the troposphere down to ground level. Comparison with MLS and Ozone sonde data show, within the resolution of the method, good agreement.

Aerosol, Surface reflectivity and Clouds

Various groups report initial success in attempts to derive aerosol amounts and classifications from the GOME data. The results of intercomparison of GOME derived cloud cover with independent data sources (ATSR-2, METEOSAT) confirms the overestimation of fractional cloud cover (see paragraph on total ozone column)). Use of the polarisation detectors by means of sub-pixel RGB indices improves on the accuracy of cloud detection. The spectral range of the GOME instrument enables interesting studies of reflectivity, albedo, and polarisation properties of surfaces and clouds.

Conclusion

The data provided by GOME instrument in general stand up to the high expectations. Further improvements and additions are necessary for the ground segment to fully exploit the potential.

Atmosphere using ATSR, PRARE and GOME data in synergy

Aerosols play a very important role in the atmosphere studies, climate effect and satellite measurement. During this session it has been demonstrated that the use of satellite data provides very useful (and unique) information about aerosol. The combined use of ATSR-2 visible and NIR and GOME allow the derive the aerosol mixture and the UV-B fluxes.

In complement, the double view capacity and the brightness temperature of ATSR-2 give the possibility to detect aerosol concentration both in terms of quantity and quality. This result was validated with volcano aerosol after the Pinatubo eruption and with Sahara dust detection.

The ATSR-2 visible and NIR channels provide the capacity of determining the optical depth, the size and shape of cirrus cloud crystals. Thanks to the stereoscopic ability of ATSR-2, it is possible to determine, with an error margin of 3 kilometres, the height of the clouds and top temperature using the thermal channel.

A more complete study of cloud detection has been made and its results have been incorporated in the operational ATSR processor, providing a more reliable SST retrieval.

A study still on-going demonstrates the interest in using ATSR-2 for deriving sea-land-atmosphere heat fluxes. This parameter is very important for climate study and again satellite data is a powerful tool at regional and global scale for measuring this flux.

PRARE, through a network of stations, gives information about the orbit determination, the ionospheric connection and the total electron content. There are plans to include PRARE in ERS-2 altimeter data and to generate and distribute PRARE data in ‘97.

Finally, a study for coupling passive and active microwave measurement has been presented. Satellites like ERS and Topex carry both radiometer and altimeter instruments. Even if the emmisivity and reflectivity are dependent, the calibration is made independently and apparently there is an incompatibility between the two calibrations. There could be a way to calibrate the altimeter using the radiometer. (The connection to be applied seems to be a linear connection of about 2db).

The satellite data appear to be very useful for atmosphere study (aerosol, cloud, ionosphere, heat flux). The combined use of instrument data (ATSR/GOME, radiometer/altimeter) seems to improve our knowledge about atmosphere features.

Ice Properties and Ice Sheet Topography

The physical interaction of the instrument signal with the ground target is being understood to a degree where applicative use of the data is feasible.

The different instruments of ERS are strengthening each others’ findings through measurement of physical phenomena by different observation techniques.

Scatterometer

• Routine derivation of sea ice parameters, free distribution to users
• Some physical phenomena can be observed over ice shelves
• predominant wind directions (tentative)
  • melt signatures
  • percolation line

Altimeter

• Calibration of signal for extraction of DEM is advancing fast, and is currently at a level where applicative use of DEM is feasible. ~1m height accuracy RMS for gentle slopes, somehow less accurate for steeper slopes, but here further improvements through pulse analysis is feasible.
• The derived DEM has numerous applications linked to glaciological studies. Science can now be done at a global ice shelf scale. New features have been identified and are being studied more closely with complementary techniques.

SAR Detected Image

• Physical understanding of interaction with snow/ice is being derived through in-situ measurements and modelling.
• Feature tracking has been established as a tool for movement monitoring, but cross calibration with optical data remains.
• Snow cover monitoring is feasible if ground cover knowledge is sufficient to allow removal from signal.

SAR Interferometry

• Discharge measurements at grounding line are radically changing the estimates of ice discharge from Greenland Ice Shelf. This may become a serious contribution from ERS to global change studies.

Sea Ice

A total of 9 papers were presented during this session. The Nansen Center, Bergen, Norway presented a report on operational use of radar data within the ICEWATCH project. Some concerns about practical constraints were expressed by the audience, but it was agreed about the importance of remote sensing data for operational ice monitoring.

This first presentation also pointed out the high value of synergistic use of RS data which was underlined by a paper from the Canadian Atmospheric/Environmental Services about sea ice studies within the Canadian ice monitoring program. This work and the following presentation about mapping of sea ice around Greenland gave some details about first experiences with RADARSAT images for sea ice research in comparison with ERS SAR.

Several case studies with more or less comprehensive in situ measurements have been presented.

Two papers came up with new techniques of using SAR data for sea ice research. Chalmers University reports about a first try to use interferometry for ice movement detection. DLR closes the session with a presentation of an interesting method to use the complex SAR data to retrieve short time (<1 sec) dynamical features of ocean waves and sea ice.

Ice Dynamics

ERS-1 and ERS-2 3 day repeat and Tandem SAR data has been successfully used through a combination of ascending and descending acquisitions to separate and measure, with high accuracy, the surface topography and the 3-D ice motion vectors for many large ice flows and glaciers using the technique of SAR interferometry. Outstanding results have been obtained for many areas in Greenland, Alaska and Canada. Surface topography agrees with laser altimetric and GPS data to less then 10m rms. Ice velocity directions are accurate to within a few degrees and speeds are as good as available in-situ data. When combined with measurements of ice thickness from radar sounding campaigns, this new information on ice dynamics allows reliable computation of ice flux and discharges to be made, that appear to be three times bigger than conventional estimates made by observing iceberg calving. Less spectacular results have been achieved for inland areas in Antarctica and Patagonia, mainly due to strong wind affects causing a loss in surface coherence due to the local summer conditions. Clear evidence of phase corruption due to a change in atmospheric conditions has been observed in the arctic. The origin of this effect is unknown, but when present it can significantly effect the accuracy of retrieved topography and ice motion. This area requires further investigation. In addition, work is in progress to model the complex process due to backscattering at depths of the order of meters in permanently dry snow zones. If successful, this approach may allow the measurement of additional snow parameters from SAR interferometry.

ERS Scatterometer has been used to generate long time-scale (yearly) sequences of normalised radar backscatter for the entire Antarctic continent. Further analysis will link with a surface map from the ERS altimeter. This time series demonstrates clearly the large-scale variations in surface snow melt conditions, with a strong snow melt in the summer of 91/92 observed. Cooler (low melt) conditions in following summers, possibly connected with the eruption of Mt Pinutubo, return to normal melt 1995/96 - 96/97.

In addition, several large icebergs have been tracked calving and circulating around Antarctica, providing valuable information on ocean currents and first measure of erosion of icebergs with time using active microwave instrument.

The high spatial resolution of information regarding topography and ice velocity provided by SAR interferometry provides an unprecedented opportunity to understand the dynamics of ice flows and glaciers allowing further insight into discharge and surge processes. Further Tandem acquisitions are strongly recommended to fully understand the evolution of ice dynamical processes.

Coastal Zones, Ocean Dynamics and Features

For coastal zone application we are now seeing the architecture of Shallow Water Bathymetry, wetland and river delta operational system in which SAR images play an important role. In shallow waters (< 30 m) where the current exceeds about 1 knot (0.5 m/s) the surface expression related to bottom topographic features such as sand banks and channels are clearly found. As such it provides a powerful data source for monitoring shallow water conditions (complemented with traditional depth soundings from ships). This information is important in navigation safety. River outflow and plume advection from major rivers are also expressed in SAR images with different characteristics during the tidal cycles. It is demonstrated that the details of these observations make it a valuable data source for validation of high resolution models of river outflow interaction with coastal currents.

SAR imaging of oceanic fronts, eddies, natural film and atmospheric boundary layer conditions and wind speed have advanced to a point where we now see more and more quantitative interpretations. In particular, wind speed estimates are derived from SAR images with very good agreement to in-situ observations suggesting promising capabilities in coastal regions where the wind field can undergo rapid changes as influenced by the land-ocean boundary. Moreover, characteristic SAR image expressions of fronts and eddies agree well with conceptual radar backscatter models suggesting that both current shear and convergence pattern can be obtained from the images. This combined with the SAR imaging of surface film (and oil spill) demonstrates new and promising retrieval capabilities for process studies in the upper ocean and atmospheric boundary layer.

Global Change

Seven papers were presented in a session which covered both regional and global aspects of climate change. Data from the ATSR, SAR, altimeter and scatterometer were used, together with additional data from TOPEX-Poseidon and NSCAT, in the analysis of climate processes.

Three papers related to the role of the Antarctic region provided significant scientific results. The first used scatterometer data (ERS and NSCAT) to measure variations in sea ice extent which was then used to characterise the pathway provided by Antarctic sea ice in connection with ENSO and monsoon events. In addition, sea ice variations were linked to the production and outflow of Antarctic bottom water and hence to thermohaline circulation. A second presentation used SAR data to characterise the extent of glaciers in regions of the Antarctic where the environment was close to the crucial point for glacier preservation. Such glaciers are extremely sensitive indicators of local warming and SAR data were found to be an effective tool for their monitoring.

Radar altimeter data were used to deduce that there was no secular variation in ice sheet elevation of the Antarctic. The climatic implications of this null trend in elevation were entirely dependent on the spatial correlation of temporal fluctuations in mass and density of the ice which require some 10, 20 or 100 year time series of measurements in order to be properly characterised. It is possible, however, that the altimeter time series can be used to reduce uncertainties in the mass balance from 400 GT/year to 85 GT/year.

Three presentations were also made regarding ATSR data.

A characterisation of surface fluxes at a number of test sites in Australia was presented and the excellent agreement of ATSR and ground based radiometers was emphasised, leading to a quantification of short wave surface flux in clear sky conditions of better than 10 W/M² broadband. The longwave component of the surface flux was more problematic however.

One other result with implications for global models concerns the use of global SST values in climate models. A presentation from UK Met.Office described an attempt to reduce the level of variability between ATSR measured skin temperatures and buoy measurements of bulk temperatures as a preliminary task prior to global conversion of all ATSR SST measurements to bulk temperatures for use in a climate database. Unfortunately, the skin to bulk temperature conversion remains unsolved.

As a conclusion, we are now starting to see the benefits of a continuous time series of ERS observations for global change analysis. It was pointed out however, that even the current time series is not sufficient and more measurements are required.

Winds

Operational Assimilation of scatterometer winds started in several European Meteorological Centres during the last two years, in particular with the availability of Three Dimensional Variational techniques which provide an appropriate solution to the ambiguity removal problem. The assimilation of scatterometer winds using this technique gives a clear improvement of the analysed and first guess surface winds, as well as in the short range forecasts, in particular in the southern hemisphere.

Four Dimensional Variational methods were shown to give better analyses and forecasts, in particular with the inclusion of scatterometer winds, as this technique gives a better vertical projection of the surface information.

Tandem ERS scatterometer assimilation shows that data coverage (only one swath) is the main limitation.

Along with this major theme, four other aspects are being developed.

Climatology:

Long term wind atlases have been generated through collocation of ERS data with other data sets such as in-situ buoy data, analysis, and other instruments (for example NSCAT).

Wind comparison:

Various methods and strategies have been developed to compare and validate different data sets, for example comparison in the spectral domain or triple collocations.

These new methods allow an estimation of errors in each observation system, which is important information for the use of the data to derive fluxes of momentum, humidity or heat, or to drive ocean waves or circulation models.

Wind direction ambiguity removal:

Variants of Variational methods have been developed to improve the ambiguity removal method which remains the main problem when the scatterometer data are directly used in an application. This problem is largely solved by manual intervention and for automatic ambiguity removal, variational techniques are most satisfactory.

This is an important subject to be investigated in order to promote the direct use of wind products.

Use of SAR images for mesoscale wind analysis

SAR images give access to small scale features in the wind structure, in particular close to the coast. This information is very useful to validate mesoscale atmospheric models, which are becoming more and more common. The correction for the derivation of wind direction remains a problem.

Waves

During recent years, operational assimilation data systems have been developed in order to allow the integration of ERS data into wave prediction models. it is proved that this integration presents overall positive results.

However, in order to get full advantage of the information contained in SAR wave products and to eliminate differences which can be identified when comparing real data with the wave models, further improvement is needed in the SAR and RA wave products themselves and in the assimilation models.

In addition, new applications are appearing, like the wave period calculation by using RA data or the analysis of long term trends in global wave climate.

The possibility of using SAR Wave data in order to retrieve wind speed information or as a high resolution scatterometer is being analysed.

All these improvements are complemented by new data exploitation techniques, such as neural networks.

Sea Surface Temperature

Six papers were presented in this session, all involving the ATSR. The first three concentrated on the use of the ATSR for describing marine processes, while the latter three were investigations of the synergy between ATSR and m-wave measurements in the retrieval of oceanographic information. One of the main results (highlighted in two papers) was the confirmation of a correlation between Sea Surface Height and Sea Surface Temperature variations using the ATSR and radar altimeter. The maximum correlation was shown to occur for areas where meso-scale activity is high, something supported by the fact that the wavelength where this maximum correlation occurs is of the order of 600-800 km.

The tandem phase allowed the joint use of the ERS SAR and ATSR instruments to observe the same ocean features with a delay of only 30 minutes. Although searches through the data archives were impeded to a certain degree by the lack of browse ATSR data frontal structures were observed a number of times in the North Atlantic by both instruments. In addition, bathymetric structure in the southern North Sea was observed in both SAR imagery and ATSR-2 visible channels. Some features were more prominent in one sensor than the other (and vice versa) suggesting there is scope for further useful work.

A further result of interest to the oceanographic community concerned the analysis of the bi-modality of the Kiroshui extension current using ATSR SST data and CTD (Conductivity, Temperature,Density) measurements. Additional SST measurements off the coast of showed that the direct current mode coincides with wind stress anomalies east of the Philippines and SST anomalies in the equatorial Pacific.

Two other papers were presented on the ATSR. The confirmed the assumption made in the processing chain that the emissivity of the sea surface remains unchanged as the wind speed increased. The effect of foam remains to be investigated.

The second paper demonstrated the possibility of using the visible channels of ATSR-2 to study phytoplankton dynamics in Lake Ontario. This paper showed that ATSR-2 does indeed have potential in this area despite the fact that the gains and wavelengths were optimised for land applications.

Several speakers had made use of the ATSR ASST product (available for research purposes on CD-ROM from Rutherford Appleton Laboratory) which is very well suited to the study of mesoscale and larger scale phenomena.

Ocean Circulation

The ocean circulation session hosted many regional applications, half of which based in the Mediterranean Sea. Most results were based on merged Topex-Poseidon and ERS altimetric data, often using AVISO's Sea Level Anomaly data set. It was clearly established that complementary data sets of at least two satellites are required to monitor ocean circulation. Low energy mesoscale features are now resolved in the Mediterranean Sea.

Altimetry

Good ERS radar altimeter data sets now exist and are being widely used. The Geophysical Data Records have reached a high level of maturity. It was stressed by the participants that they would like ESA to reprocess all the past geophysical records to the highest level of quality, thus ensuring homogeneity of the long time series of ERS-1 and ERS-2.

Finally, the work on the ERS-2 Radar Altimeter and Microwave Radiometer calibration was reported and various absolute calibration methods involving tide gauges and transponders were presented as well as some plan for ENVISAT calibration.

Marine Geoid

The availability of the ERS-1 geodetic phase data led to the full declassification of the Geosat Geodetic Mission data, making thus available a complete dataset covering the earth surface very densely.

The geoid could hence be determined with the unprecedented resolution of a few km. The gravity field anomalies have also been mapped with the same resolution, revealing features unknown so far, and the seafloor topography could also be inferred.

The detailed knowledge of the mean sea surface and the existence of a long altimetric time series have generated also a new potential application: sea level change detection, which would be a major step in the knowledge of the environment. This application requires an instrumental accuracy and stability around two orders of magnitude better than specified, but this is not unachievable.

Finally, the general feeling of the community is that the results of these studies could be of interest to other groups, and in this perspective it would be advisable to gather the various models concerning delta-g, geoid, mean sea surface and seafloor topography on a CD and make it available to the public for promotion.

Meteorology

The meteorological session consisted of six presentations and one poster. The authors have been working with all active ERS Sensors, but mainly with the scatterometer. Data assimilation aspects were not included in the session.

Most meteorological users are now exploring the possibility of using scatterometer information to explore phenomena which are missed by numerical models due to a coarse resolution. This is the case for the s airflow changes induced by mountainous islands. A methodology has been developed for New Zealand. The scatterometer can also depict katabatic winds over a polynya and surface circulation of Antarctic mesocyclones. The latter applications suffer from problems related to the dealising, which may be wrong due to erroneous analyses in data sparse areas. Other better known meteorological use of the scatterometer refer to the following of typhoons, investigations of monsoons and El Nino as well as rainfall estimate.

The work related to the use of SAR images concentrates mainly in the tropical rain cells, whose structure can be related to SAR signatures: gust from, convergence zone, downdraft area and associated rainfall.

Finally, a model has been proposed to prove the use of the ERS radar altimeter to infer rain rate as this dampens the microwave signal.

A general recommendation from this session was that a further exploitation of synergy between SAR and scatterometer was needed. The availability of NSCAT was also welcome, as it complements nicely ERS scatterometer coverage, especially in areas where SAR is frequently used.

ATSR Instrument Performance

During this session the instrument performances of ATSR were discussed.

ATSR-1, after 5 years of operation, is still able to provide data of good quality, better than AVHRR. After the anomaly of ATSR-2, from December 95 to July 96, ATSR-2 is now fully operational. All the parameters indicate a very good stability and good quality, better than ATSR-1. The on-board visible calibration system works well and in situ observations yield drift less than 1.7% per year for the visible channel.

In situ experiments, like the Mutsu Bay experiment, have allowed the validation of the SST. The results found give an accuracy of about 0.3ºC, which was initially requested by the climate user community.

Orbits

The PRARE instrument works very well, and the network of PRARE stations covers both hemispheres. PRARE shows up as a very good system for precision fast-delivery orbits.

The initial precision of the ERS-1 precise orbits was around 50cm. The orbit has now reached the unprecedented radial accuracy of 7-8 cm, due to the recent improvements in the gravity models, which were based on the altimeter measurements during the geodetic phase.

Furthermore, today’s work demonstrates the use of ERS tailored models and the use of simultaneous determination of ERS and Topex orbits through dual crossover analysis. This reduces the error to 5 cm, which is better than the basic requirement for global oceanography.

ENVISAT will strongly benefit from theses achievements in the ERS orbit determination and related gravity models.

ENVISAT Special Session

The special session dedicated to the presentation of the ENVISAT mission attracted about 200 participants and consisted of an introduction to the mission objectives, to the payload elements and their individual objectives and to the main mission operation concepts. It was concluded by a special presentation by NIRA (Norway) on the application potential of MERIS (Medium Resolution Imaging Spectrometer), one of the key instruments flying on ENVISAT.

The ENVISAT mission, to be launched in 1999, is dedicated primarily to environmental monitoring on local, regional and global scales. The complex payload allows to address issues in the domains of atmospheric chemistry, physical oceanography, marine biology, glaciology and ice processes, coastal processes, land and vegetation studies and climatology.

The ENVISAT payload will also allow the continuation of several measurement series currently provided by ERS: wind speed and wave height, wave spectra, sea surface temperature, SAR imaging and interferometry. For some of these measurements it will make available, in combination with ERS, continuous measurements over a time frame of 14-15 years.

The same payload will substantially contribute to the monitoring and management of earth resources at local and regional scales. ASAR and MERIS, with their operational flexibility will allow to prefigure future operational mission configurations in support of specific applications. The same instruments will support commercial projects. The whole payload allows to provide support to operational entities in various domains (and with different degrees of readiness): meteorology, environmental monitoring, climate monitoring, sea ice monitoring, navigation, oil pollution monitoring, etc.

The mission will provide the capability to record on board most of the data sets, ensuring a global coverage for all low rate instruments. High rate instruments (ASAR and MERIS full resolution) will be acquired by regional stations or through the Artemis data relay satellite, thus ensuring a quasi global coverage.

A well defined set of ESA products will be available through several sources, both in near real time and off-line. Co-operation with national facilities of participating states will allow a progressive enlargement of the services proposed to users.

More information is available through Internet (web site through the ESA home page, ESA programmes, Earth observation) and on a brochure distributed at the meeting and available on request from the Mission Management Office, ESA HQ, 8-10 rue Mario Nikis, 75738 Paris CEDEX 15.

Working Sessions

On top of the oral presentations and the poster presentations, working sessions were organised to discuss specific questions.

Atmospheric Artefacts in SAR interferometry

Variations in atmospheric conditions between repeat pass INSAR lead to spurious signatures in the interferogram. The dominant effects are believed to be due to variations in the tropospheric water content and in the ionospheric electron density. The phase signal is independent on the altitude of ambiguity. When fringes are converted into heights for generation of DEMs the artefacts can add considerable error to the small baseline (large ambiguity) interferogrammes. With increasing baseline the effects of atmospheric variations decrease, but the fringe unwrapping of the interferogram becomes more challenging at the same time as the common spectrum of the two acquisitions decrease, resulting in lower coherence. The understanding of the artefacts and the techniques which will allow applicative use of the Tandem dataset are still under development.

Based mainly on an analysis by ISTAR the status of atmospheric artefacts in ERS SAR interferometry today can be summarised as follows:

Some 10% of the interferograms they have processed over temperate zones (Europe, USA) contain localised artefacts (1 to 3 cycles) with spatial scale smaller than 10km.

A low amplitude signal of 1/3 to 1/2 phase cycle is generally present in all interferograms with spatial scale smaller than 50km.

There is ongoing work in trying to identify a correlation between atmospheric artefacts and tropospheric cloud inhomogeneities through use of IR and visual data collected over the site close in time to the acquisition. If a close correlation is found there is a high probability that identification of acquisitions with optimal conditions can be implemented. The generation of masks for areas with high artefact levels may be possible, and in particular cases it may be possible to derive phase corrections from the auxiliary data. The feasibility of this approach is, however, to be confirmed.

Tropospheric artefacts created by air/land interaction have been observed in interferograms. It is a known fact that topography influences the local atmospheric conditions. Examples of such effects are mountains barring the way of cloud systems and creating leeway wakes.

The presence of ionospheric artefacts should be most pronounced close to the poles and close to equator. At mid latitudes they should not contribute significantly to the artefacts.

The ionospheric phase delay can be measured by GPS ground stations. In areas where the spatial density of such stations is good enough one can generate maps of ionospheric electron density. This can be used to investigate scientifically the link between ionospheric activity and artefacts, but the generalisation of this methodology is precluded by the limited areas with high enough GPS station density.

Modelling of ionospheric activity based upon sunspot activity may be an approach to removal of the large scale ionospheric variations, but the spatial detail of such corrections will probably preclude the possibility of removing the finer variations in the ionospheric clouds.

Efforts are being undertaken to establish tools for identification of Tandem pairs with low probability of artefacts. The first approach has been to gather world climatological data and compare the probabilities for cloud cover and precipitation for each month throughout the world with the acquired Tandem dataset. Further work needs to be done to assess the potential of this technique in identifying Tandem acquisitions with low levels of atmospheric artefacts.

POLIMI presented a technique for removal of artefacts through the use of multiple interferogrammes and maximum consensus decision process has been developed. This allows the generation of artefact masks as well as quantitative measurements of the artefacts. The technique currently needs up to 6 interferograms with reasonable coherence and a range of baselines from around 100m to 300m to yield optimised results. With further development it should be possible to limit these somewhat stringent requirements to allow a larger portion of the Tandem dataset to be used in applications requiring artefact removal. The application of this technique in areas where the coherence is good after 35 days is an additional possibility. In this case one can expect larger areas to be covered with data suitable for applicative use.

During the session a range of recommendation were proposed by the participants and commented on by ESA.

A second Tandem mission of at least 6 months for acquisition of a higher baseline dataset (the current dataset has baseline in the lower range of the useful spectrum) and for the addition of multiple takes over new areas to allow removal of artefacts through the maximum consensus technique. It was mentioned that for many foreign stations, such a campaign will have to be planned up to 8 months ahead due to the station logistics.
Comment: The addition of a shorter Tandem campaign can be foreseen to fill in holes in the existing coverage. Although there are areas (e.g. in Europe and parts of Northern America) covered today with a large number (>6) of tandem pairs there will only be on average less than 2 pairs covering all land surfaces. The cost in operations, the availability of the receiving stations and lifetime shortening in obtaining 3 to 4 tandem acquisitions over the whole world appears to be prohibitive. If an additional campaign is to be launched it will have to be limited to areas where there is a real market value for a DEM. There is therefore a need to assign market based priorities prior to the planning of such a campaign.

Coherence and phase confidence maps should be made available to Value Adding Companies and users to allow to determine whether a specific area is sufficiently covered with interferometric products to allow the successful derivation of a customer product (DEM). It was suggested that the quick-look prototype demonstrated by ESRIN should be used for this purpose. This interferometric quick-look processor prototype, which was developed by POLIMI and further developed by ADVANCED COMPUTER SYSTEMS, Italy, is not yet fully operational. It has been however suggested to start up a limited demonstration operation where selected parts of the Tandem archive are processed in cooperation with VACs active in the field.

As an alternative to a new Tandem campaign a 3 day repeat cycle was proposed. This cycle, however, would not allow for global coverage at lower latitudes and would reduce the application potential of some of the Low Bit Rate instruments.

Observation of Landcover from ATSR data

The purpose of this session was to discuss the ATSR-2 potential or possible problems for study over land.

ATSR-2 presents three features, different from previous existing sensors which should help the land use community:

• Narrow band in visible-NIR: we expect to observe vegetation change with a much better sensitivity.
• Two angle views (initially designed for SST). This should help in atmospheric correction and give much new information about bi-directional effect.
• On-board visible calibration system. This provides a very accurate calibration for the visible and NIR channels.

A presentation was given about the angular dependence of radiometric surface temperature for sparse vegetation. At an incidence angle of 55º (corresponding to ATSR-2 forward view) over sparse vegetation the bi-directional emissivity effect can reach a difference of about 2º C.

The effect of atmosphere being negligible in this experiment. This study needs to be completed with more ATSR-2 data.

Using Principal Component Analysis, it has been demonstrated that, due to spectral channel definition of ATSR-2, the spectral resolution of ATSR-2 is similar to that of LANDSAT.

The 1.6 mm channel provides for distinction of organic and mineral soil types. This makes ATSR (1 and 2) a powerful tool for mapping of forests and burnt surfaces.

Over land a problem of mis-collocation between nadir and forward view has been noticed. This is caused by the relief. For precise collocation nadir/forward, the relief effect needs to be corrected.

Combining the spatial resolution of 1km (already demonstrated to be very useful by AVHRR experience) and the good spectral resolution ATSR-2 is a very good tool for land studies. Different studies or projects using ATSR-2 over land are still ongoing (heat budget over urban area, fire detection, volcano monitoring, forest mapping).

However, this promising perspective needs to be confirmed and demonstrated operationally. There is a perception by many potential users that AVHRR is easier and less expensive to obtain. This acts as a deterrent. The ATSR-2 data promotion and distribution need to be improved.

JERS/ERS Synergy

This session, arranged as a result of an ESA - NASDA cooperative working group, provided a useful opportunity for international scientific participants to this symposium to discuss progress in developing applications projects which can exploit the complementary capabilities of the currently orbiting ERS C-Band and JERS L-Band SARs.

The session was well attended and topics covered included ice-cover monitoring in climatically sensitive regions around Japan, as well as a comparative assessment of the Interferometric capabilities of the two SAR systems.

Six of the eight presentations addressed the joint use of ERS and JERS for vegetation monitoring, particularly for forestry applications.

Two papers gave a brief summary of progress on understanding and modelling the underlying microwave scattering mechanisms, with emphasis on the wavelength, polarisation and incidence-angle dependencies which determine the characteristic responses of each SAR over specific vegetation types, environmental and observation conditions. Based on this, the complementarity of the two systems, as well as their limitations, are well understood. This provides a sound basis for exploiting the operational observation capabilities of the two SARs for achievable goals within well defined applications.

In two presentations the capability to distinguish between different tree species using composite JERS/ERS imagery was reported. In a study conducted over test sites in West Malaysian plantations it proved possible to distinguish rubber trees from palm trees in composite imagery. Results obtained from a classification scheme aimed to produce vegetation maps at scale of 1:200,000 utilising JERS and ERS SARs were also reported and ability to discriminate according to a standard physiognomic vegetation classification scheme were presented for conditions ranging from desert to rain forest. Classification accuracy up to 90% was reported for certain vegetation classes with ground truth for test sites.

Two major projects aiming to exploit the global-scale potential of the ERS and JERS SARs for forestry monitoring in the coming year were presented. A campaign of nearly simultaneous ERS/JERS acquisitions over Siberian Boreal forest zone, for which the DLR transportable station will be located in Ulan Bator is planned for June 1997. A similar exercise, linked to the on-going JERS Global Rain Forest Mapping Project, will compile several SAR (ERS, RADARSAT, JERS) data sets at over the African tropical rain forests, building on the techniques and results already obtained by JRC ASPIRE within the TREES project, was reported.

In both these projects the objective is, using orbiting SAR systems, to generate semi-continental image mosaics and maps of vegetation parameters over climatically sensitive forest regions for which no other comparable data set is available.

These presentations highlighted the usefulness and complementarity of the ERS and JERS SAR missions. Participants from ESA and NASDA acknowledged the benefits of this type of scientific exchange in view of the forthcoming C-band and L-band SAR missions currently being prepared by the Agencies. NASDA representatives expressed their thanks to ESA for arranging this session.

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