Webový obsah - zobrazení

Assessing flood risks with severe lack of data, Reni case, Ukraine

The aims of the project were 1) share expertise about flood risk mapping, emergency management and damage calculation 2) create better flood prevention and protection plans for the Ukrainian part of the Danube Delta 3) contribute to the set up of an integrated flood risk management plan in line with the WFD and EU Flood Risk Directive 4) assist with capacity building and community involvement to flood risk management and 5) formulate recommendations for data collection.


The city of Reni is under threat of flooding from the Danube as well as from flash flooding due to extreme rainfall from a smaller river which runs through Reni. The area of Reni was chosen as a case study for several reasons. It is a community where during a flood not only 8000 population and housing are at risk but also a harbor and a water intake for water supply. Further the community and the water managers over there, together with the Danube Flood Protection Department, the Emergency services, the Town of Reni and the Centre for regional studies in Odessa were all giving their full support to this project.

Normally state of the art models are used to execute a FRA. These require however al lot of (detailed) data. In the Reni case there was a severe lack of detailed data. By involving the right stakeholders using their local knowledge and making the FRA a joint effort useful results can be produced and can be the start for a more effective flood risk management. In this way a flood risk strategy can be formed without having all the data beforehand.

In earlier days only mitigation ( taking measures like dams, dikes, …) was done to limit flood damage. Nowadays, flood risk manage comprises a lot more and developed towards a whole chain: before, during and after the flood. ( see figure 1). Very important to be able to make the right decisions during all these steps is the flood risk assessment (FRA).

Figure 1. The different steps in flood risk management

Reni case study

Reni is a small town in the Odessa Oblast of Western Ukraine figure 2). The risk of flooding comes from two sources: Flood damage is caused at one hand by high water level in the Danube, with associated risk of dike breaches and on the other hand, as an extra threat, flash floods due to heavy rainfall and runoff from the upstream Balaneshty hilly catchment, just upstream of Reni.

This combination of two potential flood sources leads to an even more severe level of flood risk (higher flood probability and higher impact !) As a transboundary part of the Danube delta, some extra management issues are to be taken into account by developing a proper flood risk management along this Ukrain-Romanian Danube delta stretch.

Figure 2 shows the area of Reni. The red and yellow lines on this map indicate a flood extent in of a major flood event in 2005 from two different sources.

For the moment there is not a comprehensive flood risk management plan available. When a flash flood happens, or when a river bank along the Danube seems to collapse, the Department of Emergency Planning, helps where they can by reinforcing dikes with sand, and by helping and rescuing people who are trapped by the flash flood. Actually, in the flood safety chain only some very limited actions are taken at the pro-active side while most of the flood risk management is situated in the response phase (after the flood event). After the severe flooding of 2005, a large number of houses and infrastructure were completely destroyed and a lot of Reni inhabitants were stuck in their flooded city.

Figure 2. Threats to flooding and delineation of the 2005 flood extent


The flood extent (in space and time) and associated flood depth is determined for a real event in the past, but also for different synthetic storms with different statistically determined return periods. These floodings are visualised on GIS maps.

To determine the flood extent and flood depth the flash floods are numerically modelled. For this case study, this hydrological-hydrodynamic schematization of the river environment was done in InfoWorks RS software. InfoWorks RS includes full solution modelling of open channels, floodplains, embankments and hydraulic structures. Rainfall-runoff simulation is available using both event based methods and conceptual hydrological methods. Full interactive views of data are available using geographical plan views, sectional view, long sections, spreadsheet and time varying graphical data. The underlying data can be accessed from any graphical or geographical view.

However, main obstacle for the creation of the flood maps are data. We will show here that even with a small amount of data, it is possible to create useful maps.

The modelling approach towards the computation of flood maps relies on the set up of different numerical models interacting together (Huygens et. al., 2009). The models are calibrated at the hand of available historical data. The following models were set up: Digital Terrain Model (DTM), Rainfall-Runoff (RR), Hydrodynamic model (HM) and Flood mapping (FMM).

The key issue on flood mapping is the representation of flooded areas related to a certain statistically determined return period. The procedure above ends up providing a tool for transforming rainfall into flooded areas. A final step comprises the statistical analysis of several flood events in order to derive the frequency of occurrence. Therefore an Extreme Value Analysis on the results at each model node for multiple simulations is carried out. For this purpose a synthetic set of rainfall events was generated to mimic the available characterization of rainfall parameters. Each of these events was simulated with the hydrodynamic model ending up in a cloud of stage and discharge values on each one of the 125 cross sections. At each location an extreme value function was fitted for stages; subsequently the fitted function was used to generate stages for several return periods. Stages for corresponding return periods lead finally to the required flood maps through the flood mapping modelling.

Data availability

Data for the analysis was mainly provided by local authorities and stakeholders within the region. As a general characteristic it was scarce while resolution and reliability were limited. The following data was available:

  • Topographic surveys: A limited number of cross sections were surveyed and documented in the report about the 2005 flash flood. This data, originally in analogue format, was digitized for subsequent use.
  • Structures: The location of 11 bridges along the Balaneshty river reach where known information on the flow openings under the bridges was limited; surveyed dimensions were not available. There is no evidence of water control systems (gates or weirs) with manual or automatic control.
  • Satellite imagery: Satellite imagery becomes important when available surveyed data is not sufficient to describe topographical features of the terrain with sufficient accuracy. For Reni 3 different sources were used:
    • ASTER Global Digital Elevation Model; (ASTER GDEM), a DEM data which is acquired by a satellite-borne sensor "ASTER" to cover all the land on earth with horizontal resolution of up to 30 m.(http://www.gdem.aster.ersdac.or.jp/)
    • SRTM digital elevation data, produced by NASA. Through the CGIAR-CSI GeoPortal SRTM 90m Digital Elevation Data for the entire world is available. (http://srtm.csi.cgiar.org/)
    • Reni Satellite image with 5 m horizontal resolution.
  • Climatological data: Climatological data for Reni was limited to short time series of rainfall and evaporation for July 2005 gauged at stations Reni and Nagornoe. For the stations Bolgrad, Izmail and Vilkobo, located outside the study catchment, a long time characterization of rainfall parameters limited to monthly parameters (maxima, minima and average) for a gauging period between 1945 and 2005 was also available. Water levels gauged at the Danube in Reni and at the Kagul lake were also reviewed. Flow discharge data was not available.
  • Historical floods: This is a key input in order to calibrate the numerical models allowing them to produce realistic outcomes. An analogue plan containing a delineation of zones that were flooded in July 2005 was the only historical information that could be used. Personal observations by some privileged stakeholders have helped us to reconstruct the flood picture of the 2005 event, both in a qualitative and quantitative way.
  • Site visit and photo report: During a site visit to Reni the Balaneshty river was photographed. At the hand of this photo-report a general idea of the conveyance capacity of both the river cross section and the bridges could be derived based on expert judgement.


We created with the model and the available data a lot of maps of the region. After we created the flood risk maps, we used the maps to enhances both technical training and operational capacity building. Further we organised an infomarket. In the morning, specialists in flood management (modellers, emergency planners, water managers and policy) were invited to discuss the results of the model for Reni. It was discussed how a better flood risk management plan could be set up with the aid of the newly created information and maps. In the afternoon, the infomarket was made for the public. During the whole afternoon the people could walk in and see the results of the project, maps were posted against the walls. There they could search for their house, the possible risk and look for an evacuation spot. Different experts were present to help the people with there questions and to hear there ideas. The whole idea behind the info market is:

  • to focus on methodology rather than on technical details, making the steps in flood risk mapping easier to follow for 'non experts';
  • to provide one with results to discuss with the stakeholders, allowing them to actively take part in ánd see the benefits of flood risk mapping;
  • to actively involve the public and make them aware if the flood risk in their to help to identify the "gaps" in the data in a relatively early stage making the stakeholders aware of what data is needed for flood risk mapping and what effort is required to acquire that necessary data. it allows one to focus on acquiring thát data which will help to make the most effective flood risk management measures.

In the next figure 3 produced flood risk maps are presented with different return periods. The next figure 4 is a flood risk map for the 2005 flash flood event, together with indication of critical points, bridges, evacuation routes, gathering spots. Together with this map a house is shown with indication of the flood depth as indicated with colours on the maps. That makes it very visible and understandable for the public what will happen.

Figure 3. Flood risk maps with different return periods: 5, 10, 20 and 50 years

Figure 4. Flood risk map of the July 2005 flash flood event with indication of evacuation shelters, bridges, water points, …

By showing important points on the maps, people can see very clear what will happen during a flood. That might be a bridge that one will not be able to use anymore or they can see that a possible shelter place, like a cultural centre will be flooded with less than 20 cm. For emergency planners these maps will also reveal a lot of extra information to set up an emergency plan. It will for example show if drinkwater inlets will be flooded or what kind of evacuation routes must be communicated to the people living in flood prone areas, etc.

Recommendations for data to create more accurate maps and plans

  • Topographic surveys: The geometry of the conveying channels is currently not sufficiently accurate; the broad irregular bedding may need to be topographically surveyed at intervals of at least 200 m. and extending at least 20 m beyond the actual embankments. The extent of frequently flooded zones is well known; surveys within this zones should be sufficient to delineate the flooding patterns.
  • Structures: All the structures, such as bridges and culverts, need to be surveyed more systematically and more detailed. Important at this point is a clear idea of the conveying areas (width, depth, high, etc). For structures, which may be overtopped (culverts, dikes, embankments, etc), flooding thresholds become important. When relevant for urban drainage/sewer systems discharging into the main channel, locations and weir dimensions will be necessary.
  • Climatological data: Rainfall, evaporation, water levels and river discharges are required parameters for the numerical hydrological and hydrodynamic model development. This is currently a major gap in the simplified approach. For each one of these parameters there are two types of datasets that need to be obtained: long time series and historic events. Long time series within the catchment can not be produced any more the alternative is obtaining this sets for neighbouring locations or locations displaying similar characteristics but further away from the study area. A proper data registration for events in the future can be gauged relying on the implementation of a new gauging network comprising at least a couple of stations where water levels and water flow are recorded in a systematic and continuous way. In recent years, some climatological data, obtained by satellite imagery, has become available for certain parameters like rainfall, temperature and evaporation. The most important advantage of this source (radar spots) is the spatial coverage and variability of the data; less attractive is the time resolution (3 to 9 hours sometimes) which can be a disadvantage for small catchments. For this kind of global spatial data ground thruthing with local observations is strictly needed to have a more reliable calibrated dataset.
  • Historical floods: Historical characterisation is a practical alternative when gauged data is insufficient. Based on bibliographic research and interviews with local authorities and stakeholders it is possible to qualitatively reconstruct important flood events in the past. Depending on the sources, even a limited quantitative character can be achieved. The characterisation can be time consuming but it usually delivers very enriching input for both modelling and management.
  • Economic data: A database with economic data about value of properties, landuse, railways, roads, harbour, water intake point should be available to calculate possible economic damage.

Conclusions and recommendations

This project showed how already flood risk maps can be created and be useful even without a lot of data or detailed data. This project could start a lot of activities like expertise building around flood risk mapping, set-up of emergency plans and public awareness raising. All this was done on case study level in the city of Reni. Now the results should be used to start these activities in the whole Ukrainian part of the Danube Delta. Also in Reni still a continuation of the started activities is needed. Not the least the collection of more data and the upgrade of their existing data. Upgrading data sources implies the need of upgrading, enhancing and recalibrating all the numerical models. Based on new surveyed data it will certainly be possible to produce a new DTM fitting better the actual topography of the main channel and flood plains. With adequate climatological data it will be feasible to develop a set of conceptual RR models. PDM seems a good alternative to investigate. New survey data and RR models will lead to a new conceptualisation of the hydraulic and flood mapping models. The all around result would be a more accurate and reliable set of flood maps which provide a better understanding of the risk and allows for effective measures (both in effectiveness and cost efficiency).operational flood forecasting and warning system.