Flood Monitoring Workshop Report

G. Calabresi

ESA/ESRIN, Earth Remote-Sensing Data Utilisation, Frascati, Italy

The Tanaro map river, near the town of Alessandria, Piemonte, Northern Italy The flood which occurred on 4, 5 and 6 November 1994 was the most studied natural event in recent years. It has been, indeed, one of the worst disasters ever happened in this region during the 20th century. The map section contains GIS information (raster based) combined with Spot panchromatic and ERS-1 SAR data interpretation. As a good example of multi-sensor use, the Spot image interpretation made from the panchromatic data collected only 17 days after the event, allowed to map the entire damaged area by means of the sensor response to the fine material sedimented during the flood. The availability of ERS-1 SAR data in near real-time (9 November) allowed, instead, to monitor the flood event soon after its occurrence. The inundated areas (well visible in green) correspond to paleo-meanders of the Tanaro. (Courtesy of CSI - Consorzio per il Sistema Informativo, Piemonte, Italy)

GENERAL COMMENTS

The organisation of Working Group meetings of this nature is quite new in ESA. It was inspired by the idea that the objectives of ESA's Earth Observation Programmes have to be pursued not only in the scientific but also in the application and operational/commercial domains.

The choice of the 'Flood Monitoring' theme for the first meeting was based on the identification of flooding as one of the most promising applications of SAR. Useful flood assessment maps can be generated using SAR data information. In addition, because of the presence of clouds and bad weather conditions in any flood event, SAR is often the only available tool in the hands of decision makers, civil protection authorities and similar, during the monitoring phase as well as the post-assessment phase, when damage evaluation exercises have to be carried out.

In compliance with the background and scope of the meeting, i.e. to bring together parties interested in the specific theme, the presentations were grouped under two categories:

• Providers of satellite information and
• Users of information.

Aspects related to flood event monitoring, assessment and, in some instances, prevention, were treated with more emphasis on the application of remote-sensing data, results, advantages and disadvantages, effective use of final value-added products, than on 'technicalities' such as algorithms, image processing techniques and methodologies.

The integration of optical/microwave data with GIS information was applied in several projects and lead to good results in the majority of them. It is considered as an important step towards the operational use of remote sensing also for flood monitoring.

A crucial point was identified in the transfer of project/study results to the end-users, which proved to be, in most cases, not easy and even sometimes impossible to realise, due to constraints of various kinds, such as user degree of involvement, full understanding of end-user requirements, proper information tools...

Once more, a certain lack of confidence in the SAR instrumentation was felt. Two main factors, such as the still poor degree of familiarisation and the opinion that the technology is somehow difficult to handle, seem to prevent its wider exploitation.

THE INFORMATION PROVIDERS

Presentations by this group indicated that the projects/studies were carried out within three main frameworks of activity:

1. research
2. demonstration and research
3. technical.

The most common output, in general, were maps with flood extent delineation. The maps were obtained using Spot, airborne SAR, ERS/SAR, Landsat-TM, SIR-C, alone or in combination. They all referred to flood events occurred in Europe in the last three years.

ERS/SAR was fundamental in the case of Piemonte (Italy) floods, being recognised as the unique source of information available. It allowed to generate a valuable 'working tool' in the form an up-to- date map for immediate use by the authorities, who were thus able to derive a first assessment, both of order of magnitude and of damages. The map is integrated into the regional geographic information system (GIS), can be used in further studies or projects for evaluation of agricultural damages, for comparison between today's situation and the town expansion in flooded areas during the past century, as well as for land-use planning and monitoring activities in general.

The good coordination of information producers and civil protection users was demonstrated in the report on a flood relief and monitoring project in the Arno River Basin (Tuscany region, Italy). SAR was used to obtain (a) an inventory of the flooded areas by the visual image analysis, and (b) a flood extent prediction by means of signal variation analysis. Consequently, assessments of damages to crops and structures were possible. A high interest in such good results was shown by the Pisa Province authorities, who declared their intention to sign a convention for a flood monitoring and relief project to be operated in their area.

Three case studies in different regions of France provided good indications of SAR remote-sensing usefulness for flood monitoring. A conclusion was that if an inundation has to be precisely located, geographic information plus remote-sensing data, together with the required data integration tools (GIS), are absolutely necessary.

The results achieved in Germany in several projects combining X-SAR, Landsat-TM, ERS/SAR and aircraft SAR data, are encouraging. The importance of ERS was emphasised, even if aircraft data and DEM's are recognised as essential for detailed mapping. In some instances, data acquisition constraints and also some scepticism shown by the end-users about map accuracy were, unfortunately, in conjunction with budget limitations, the cause of project interruption or suspension.

Very promising perspectives were disclosed by a research study in the UK, aimed at the development of a high-resolution model for flood forecasting. The role of SAR, so far, is model validation. However, it is suggested to use SAR data to refine the necessary DTM's and to define a realistic scenario of the flood extent for initiating the model computation.

As it had been anticipated during the first ERS-1 Pilot Projects Workshop (Toledo, Spain, June 1994), the experiment conducted in Belgium at University level to characterise flooded areas and to delineate poor internal drainage soils using ERS-1 SAR imagery had a very interesting follow-on. The end- user, the Walloon Ministry of Equipment and Transportation, signed a contract with the University for the provision of updated maps of inundated areas along the main rivers visible in the ERS-1 SAR images of Wallonia. A further step to achieve a more satisfactory transfer of technology from the University to the Ministry will be a cost-benefit analysis planned in the near future.

The preliminary scheme of a Telenetwork Project called 'Floodnet', elaborated in the UK, was presented at the meeting. Still in a pre-design phase, Floodnet aims at increasing the user's awareness of Earth-observation potential by the development of a EU-wide communications network.

THE INFORMATION USERS

Three main types of institutions/companies were represented at the Workshop:

• potential users
• scientific/technical users
• insurance, re-insurance companies.

The presentations were good opportunities to compare opinions, views, needs, suggestions. However, the scenario in which producers and users operate still reveals critical areas, such as:

• not enough dialogue between the parties
• lack of correspondence between offer and demand
• need of a deeper cost-benefit analysis
• unavailability of on-line low-resolution SAR services (i.e. real-time services with adequate repeat frequency).

The important role of national space agencies in funding projects carried out in cooperation with either value-adding companies or scientific/technical institutions, or both, was put into evidence. The addressees of such projects output are primarily regional authorities, ministries of environment, etc. These, however, often need to be convinced about the maps accuracy and express doubts on flood monitoring efficiency, where ERS-1 repeat cycle is considered as inadequate.

It was interesting to learn how flood events are managed and controlled in Holland, where satisfactory results are achieved by an efficient network of 50 MS (Measuring Stations), and by daily prediction whenever potential flood situations occur. The information collected is complemented by aerial photography data. Satellite remote sensing is considered suitable only to large spatio-temporal scale studies.

A quite original application of satellite radar data was related to the estimation of discharge from braided rivers in British Columbia. The clear, interesting message was: the understanding of hydrologic regimes in remote areas can be easily improved by using ERS/SAR data and appropriate discharge estimation techniques.

The presence of insurance/re-insurance companies in the user category offered the audience new perspectives in satellite information as a tool for risk management in general, and flood risk in particular. It appeared that knowing the risk of inundation is for re-insurance companies more important than being aware of the real flood extent. Study activities in different natural hazards (oceans, atmosphere, ground) are ongoing. In such issues, the need is for the best available information from any source, including satellites, to be integrated into GIS. The remote-sensing data potential is good, especially if on-line data dissemination systems are available.

THE INTERNATIONAL ORGANISATIONS

It was useful to learn that the Council of Europe's coordination role in risk management and assessment follows a user-oriented approach. Among the planned actions and initiatives, a cost/benefit analysis has to be performed by March 1996, together with the identification of information needs from decision makers in the field of risk management (which should have been completed in Summer 1995).

The demonstration projects being conducted with the aid of space technology should bring a good contribution.

At the European Union's level, the 4th Framework Programme of Research and Technological Development includes natural risks issues. It ensures the exchange of expertise between the Member States experts and the Programme itself. A support function is performed by the Programme's Committees of Experts in natural risk management matters.

A direct link with the media has been established by Eurimage through its 'Earth Watching' Project, aimed at providing within 24 hours SAR images from low-resolution data acquisition, via electronic data network.

CONCLUSIONS

An encouraging amount of flood maps were generated using remote- sensing data in general, and SAR information in particular. Through the Open Floor discussions at the end of the Workshop, it is possible to extract some important ideas, mainly:

• Importance of cost-benefit analysis to promote the use of remote sensing data.
• Selection of the most appropriate approach towards the end-user and analysis of his real needs.
• Opportunity to enlarge the Workshops participation to include decision-makers.
• Availability of ready-to-use information for immediate integration into GIS.
• Need of on-line data dissemination services for real-time operations.
• Need to fill the gap of missing technology transfer between the information producers and the users.
• Availability of more 'aggressive' information material for use by value-adding companies
• More frequent thematic meetings such as the 'Flood Monitoring' Workshop.

Note: The proceedings of the Flood Monitoring Workshop may be obtained from ERS Help Desk, ESRIN. Tel: +39-6-94 18 06 00 Fax: +39-6-94 18 05 10

Spacemap of flooded areas around Avignon, Rhône Valley, France
This multitemporal map is a combination of Landsat-TM data (July 1993) used for vegetation cartography, and of 10-m Spot panchromatic images (February 1994) used for the determination of built-up areas and infrastructures. To this map the flooded areas of 12 and 15 January 1994 (yellow hatching) and the areas under water of 16 October 1993 (red hatching) have been superimposed. The data are taken from two coloured compositions produced from ERS-1 images. The multitemporal approach requires that one of the dates chosen must be during or very close after the flood maximum. It is then possible, in combining the characteristic response of the zones under water in the SAR images with one or two other images acquired before or after the event, to map the land under water at the time of acquisition of the image which shows the rise of the water.

The digitisation of the flooded areas on aerial photographs of 9-10 October 1993 provides a good complement to ERS-1 data. (Courtesy of Geoimage; contact: Sandrine Delmeire, 80 route des Lucioles, 06560 Valbonne, France. Tel: +33-9300 4000. Fax: ...9300 4001).