Figure 4-28. AVHRR image from 17/02/98 (left), and SAR image from 12/02/98 (right); centre: February statistics on Rhone plume wind drift effect, based on AVHRR (Wald, 1987).
The procedure to routinely monitor the Baltic Sea with the aid of satellite data has provided a good overview of some anthropogenically induced pollution, as well as more natural environmental events. Within the Baltic Sea test site several achievements have been accomplished during the project time. Good statistical information about oil spills, their localisation in both time and place, with the aid of SAR data is one such accomplishment. The procedure to detect and monitor algal blooms by routine surveillance using NOAA/AVHRR, SeaWiFS and in some cases SAR data is also a significant progress step. With the selection of the primary test areas, defined by 15 SAR frames, a good coverage of the Baltic Sea have been reached on both a spatial as well as temporal scale.
The use of a local receiving system for AVHRR and SeaWiFS data has proven to be an efficient way for fast access and processing of large amounts of satellite data. One data intensive event is the regular occurring cyanobacterial summer blooms in the Baltic which has effectively been monitored in near real-time during the project time. During 1997 two large-scale events occurred within the Baltic Sea drainage basin. In northern Germany and Poland a major flooding with subsequent outflow of sediment loaded river water from the Oder, Wisla and other rivers occurred. This flooding was monitored by using consecutive AVHRR data The other large-scale environmental occurrence was the ‘record high’ cyanobacterial bloom, covering more than 100 000 km2 during the summer months. Unfortunately the bloom occurred before the launch and operation of the SeaWiFS instrument, thereby no ocean colour data exist for the July-August period. For another, unusually early cyanobacterial bloom in 1997, some OCTS data exist just prior the failure of the ADEOS satellite.
The main conclusion for the work in the Baltic Sea is the usefulness of a combination of different satellite data sources such as the wide-swath AVHRR and WiFS together with fine resolution data from the TM and SAR sensors. The combination of such sensors enables a detailed view of an ongoing algal bloom in different spatial and spectral scales. The AVHRR, and today also SeaWiFS, provides excellent temporal coverage with more than daily imagery during cloud-free conditions. The SeaWiFS with its dedicated ocean colour concept also enables a much better discrimination of various water constituents. The spatial resolution of 188 meters for the WiFS sensor provides an intermediate scale when observing the algal blooms. As the first results shows from the Multi-sensor Algal Bloom Monitoring (section 5.1) this intermediate resolution is more similar to the TM data than to the AVHRR data when observing various spatial structures. Thereby WiFS, or other data with similar resolution and swath, is a good compromise between detailed view of the cyanobacterial extent (including near-shore exposure), and frequent coverage. SeaWiFS data have been routinely used for monitoring cyanobacterial blooms during July-August 1999, and in directing in situ measurements. The difference in image content and possibilities for information extraction (atmospheric conditions, pigment concentrations and so forth) is obvious when compared with the AVHRR data sets previously used.
The future with access to the "traditional" data sets from AVHRR, Landsat TM, SeaWiFS, and SAR among others, together with the new generation sensors such as MERIS, MODIS, MISR, OCM, GLI, ASAR, AATSR etc. is indeed intriguing and promising. Naturally the combination of all those sensors will require further research and development of methods to fully utilise them for operational use. During the research and development phase for the new sensor, nothing prevents the implementation of the already existing instrument systems into different operational schemes.
The studies carried out in the North Sea test region have had some notable successes in attempting to model the dispersion of sediments from the Rhine plume. The studies have also highlighted some of the potential pitfalls of using remote sensed data in temperate climate regions and of reliance on external data providers where there are limited data sources. The use of a model to assimilate all the available data is seen as an essential tool in such an area.
In order to make best use of the data for supposedly "ideal" conditions, we needed to look at test periods where we had the maximum number of data sets available over a period of at least a week, to allow evolution of features. Within two years of data, only two such periods could be found, due to the high amount of cloud cover, limiting infrared and visible imagery, and strong winds, limiting the ability of SAR to image oceanographic features. However, during the test periods studied, there was good correlation between the data sources, e.g. between SST and SAR images for 11 August.
One of the main thrusts of the research for the North Sea was to utilise the model, developed at ACRI, to attempt to recreate sediment distributions from the Rhine plume. Sediment distribution can be determined from visible wavelength sensors, preferably ocean colour sensors such as OCTS, MOS and SeaWiFS, but the visible wavebands from ATSR and AVHRR can also be used. There were severe delays in receiving the ATSR data from ESA. As the ATSR only has (at most) one pass per day, the cloud limitations on ATSR are more restrictive than those on AVHRR, with several "chances" of getting a cloud free image every day and so ATSR was being used as the limiting data source. It was initially hoped to use the enhanced accuracy of the ATSR instrument to initialise the model run, but the low number of available passes made this impossible. Recent improvements in the ATSR processing facilities, including an excellent near real time facility, should make a big change to the data situation for ATSR.
The sediment load would have been most readily determined from ocean colour data. Although these were limited due to satellite failure and delays in distribution form OCTS and SeaWiFS, we were still able to use the AVHRR instrument as a surrogate ocean colour instrument for the high sediment loads of this area. The predicted sediment distribution from the model shows many of the patterns evident in the AVHRR visible channels, and the single available MOS image. As the model does not include re-suspension of existing sediments, or the effects of wave action, this indicates that a large amount of the sediment visible in the images is directly sourced from the Rhine. This has important consequences for possible pollution incidents within the Rhine and their impact on the Dutch coast.
The time series of AVHRR and ATSR images used for this study further highlight the problems encountered with cloud cover. This time period was one of the least cloudy during the entire two years of the project, but even here, following features from one image to the next is almost impossible due to the patchy nature of the cloud cover. Use of the model allows a much more complete picture of the temperature distribution and enables time dependant hydrographic effects, such as tidal effects, to be taken into account when comparing images acquired a few hours apart.
The main conclusion from the North Sea region is the importance of using a model to provide a method of alleviating the problems of using remote sensed data for temperate coastal climates, where both cloud cover and wind speed act to restrict the amount of useful data. Future missions with coincident infrared, visible and active microwave data will improve the options for combining data, but , but using the model we will still be able to see a more complete picture of the way the Rhine plume disperses over time than will be evident from the remote sensed data alone.
The study for the Gulf of Lion Test site has lead to significant results in the expected domains. The general study of the dynamics of surface water masses has been achieved through the realisation of an extensive database, extended to the full 1995-1996 years in order to fill — partly — the gap existing between this project and the preceding ones. Original methods for an objective mapping of water masses — based on SST maps have been developed, tested and applied in order to enrich the database as a GIS where the main dynamical features are detected and tracked. The main frontal structures, i.e. North Balearic and Catalan front, the Ligurian front, the Rhône river plume (see Figure 4-28) and the limits of the deep water formation area have thus been followed during 1996 and 1997 falls. Displacements of the surface waters associated with the observed instabilities (fronts, meandering) have been evaluated. The observed high speeds have been communicated to the French and Spanish civilian authorities, which have noticed that this result give the first explanation for the high number of ship traffic incidents in the concerned areas. This information has been recognised to be of a peculiar interest for them.
The perspectives in this domain are both scientific and economic. Scientifically, the project has complemented a study started by the oceanographic community 30 years ago. The continuity of both experiments and data collecting have been exploited and extended in order to develop theoretical approaches which will lead to a proper modelling of the water masses dynamics in the area. From an economical point of view, there is now a sufficient evidence for the need to develop a dedicated analysis and forecasting service, for the use of ship companies, and civilian authorities. Such services already exist in the USA, where ship routing is optimised using analyses based on AVHRR data. The Clean Seas GIS and web-site developed in this study could be the base for a further development of an operational tool, and a market study.
The SAR coverage of the area has also lead to significant results:
(i) extensive statistics on oil spilling have been elaborated. Spills maps were revealed to be in complete agreement with the ship routes maps collected in the preliminary part of the project. According to the opinion of the French civilian authorities (Customs / POLMAR Airplane team head), the reliability of these maps is good, bringing to evidence typical reactions of ship crews: one clearly see on the maps the fact that there is no ship spill within 20 km of the French coast, because of the possibility of being identified by the POLMAR airplane, in Spain where there is no such an equipment, the situation is completely different and ships appear to spill much more closer to the coast.
(ii) the type of routes identified suggest that most types of ships are concerned with the activity of spilling, not only tankers
(iii) due to the difficulty of identification — and prosecution - far offshore when spilling, most of crews operate daily.
Figure 4-28. AVHRR image from 17/02/98 (left), and SAR image from 12/02/98 (right); centre: February statistics on Rhone plume wind drift effect, based on AVHRR (Wald, 1987).
The previous results have been partly used to operate a simulation of a pollution scenario. The hydrodynamic modelling of an oil slick, drifting to the Barcelona shore under the action of the wind has been achieved, as a demonstration case. Infrequent coverage from the SAR system does not allow acquisition successive images of a same spill. Further demonstration studies should make use of improved instruments like ASAR’s. The incidence of specific processes (diffusion, hydrodynamics, wind drift, etc.) could be taken into account.