Computer-Aided Modelling of Waterbodies
Analysis of the Hydraulic Effects of a Dike Relocation
It has been demonstrated on the Elbe near the town of Lenzen that numerical modelling prior to the start of construction work can accurately predict subsequent naturally occurring events.
Since the 1990s a dike relocation has been planned and carried out by the German state of Brandenburg on the river Elbe near the town of Lenzen. The Federal Waterways Engineering and Research Institute (BAW) supported the realization of the project with the help of hydraulic-morphological modelling analyses. The plan envisaged on the one hand changing the course of the elevated Elbe flood dike and reducing the flood channels in the floodplain with the dike relocation in the Lenzen-Wustrower Elbe Valley Lowlands, west of the town of Wittenberge. On the other hand, it included a modification to the location and structure of the floodplain forest plantings in the region of the dike relocation, as well as additional floodplain planting on the Lütkenwisch and Mödlich river islands, adjacent to the dike relocation area. Before the start of the project, in November 2006, the project sponsor, the German Federal Agency for Nature Conservation (BfN), requested intra-governmental assistance from the BAW in the investigation of the hydraulic consequences of these measures.
The BAW scientists used the hydronumerical model UnTRIM for their investigations and generated a two-dimensional model of the area under investigation. Following completion of the dike relocation at the end of 2009, it was possible to validate the model calculations consisting of changes in water level and volume of water flowing into and out of the area of dike relocation. This was performed by means of comparative measurements - water level and flow measurement - during the Elbe floods in March 2010, October 2010 and January 2011. 'It demonstrated that we were able to predict the natural hydraulic conditions that actually occurred accurately using the computer model', said Matthias Alexy, Dipl- Ing. a BAW engineer in the department responsible for hydraulic engineering in inland areas.
The two-dimensional depth-averaged hydronumerical model
Since secondary flow effects were assumed negligible in the specific floodplain and dike relocation areas under investigation of the Lenzen-Wustrower Elbe Valley Lowlands, the BAW engineers opted for a two-dimensional depth-averaged model based on the hydronumerical model UnTRIM.
The BAW model geometry of the Elbe section under consideration was largely based on a digital terrain model of the Elbe watercourse (DGM-W Elbe) provided by the German Federal Institute of Hydrology (BfG). Geometrical data were mainly taken from groyne registries, digital federal waterways maps, geo-referenced aerial photographs. The location and structure of existing and planned plantations, in particular the new dike protection plantations and a new floodplain forest, were required in order to factor in the flow resistance that they would cause. Hydraulic variants were simulated for a number of annual discharge series based on hydraulic records from the years 1971 - 2000, as well as preliminary key values of flood events from the state of Sachsen-Anhalt for the Wittenberge/Elbe gauge. In order to calibrate and validate the model, comparison was made with most recent on-site water level measurements. As the model works with stationary flow conditions, flood level data with relatively low flow fluctuations were selected.
The construction of the numerical grid for the Elbe section in question (El-km 470 - 489.5) envisaged a total of two high resolution meshes. The first represented the current state with the existing course of the dyke line at the start of the project. The second depicted the altered state with relocated flood protection dyke, the flood channels in the floodplain, the cutoffs in the Gandower Fährdamm (ferry levee), as well as the apertures (in total six openings) in the old dyke.
'We managed to represent the groyne bodies with a high grid resolution in the area of the river bed', said Matthias Alexy, 'and that with square elements with sides of 2m, such that the hydraulic effect of the groynes was captured without our having to specifically align the element edges to the geometry of the structures'. The BAW engineers likewise largely avoided having to orient the grid edges to terrain structures in the floodplain, so as to achieve an efficient mesh generation. Square elements with sides of 2m were also integrated into the model in these areas, in order to nonetheless capture the hydraulic effects of flood channels, cutoffs and dyke openings.
Figure 1 depicts the dyke relocation area (El-km 476.7 - 483.7) as a section of the overall model. In addition to the existing six specified apertures in the flood protection dyke, figure 1 depicts the Gandower Fährdamm (ferry levee) coursing right across the dyke relocation area. Three cutoffs ensure that this dam can let through sufficient water in the event of a flood. In addition to the eastern flood channel that ensures a connection to the central and southern ferry levee cutoffs, three further flood channels were constructed for the improvement of the discharge conditions in the area of the dyke relocation.
Thanks to the cuts made in the old flood protection dyke in 2009, the floodplain bounded by the new flood dyke between El-km 476.7 and 483.7 carries a substantial part of the total Elbe discharge. In order to simulate the flow resistance invoked by the existing vegetation, corresponding roughness values typical of the existing vegetation were integrated into the calculation model. Figure 2 shows the computed water levels along the river axis for conditions with and without the dyke relocation for a hundred year flood (HQ100 = 4020 m3/s, occurring on average once every 100 years).
Due to the enlargement of the discharge areas resulting from the dyke relocation, there is a significant drop in water levels in the case of flood events. 'For a hundred year flood with a discharge of 4020 m3/s, the drop in water level amounts to a maximum of 0.35m at the first floodplain inlet at El-km 477.3, which results in a decrease in water level of 0.2m at the upstream boundary of the model (El-km 470.0)', as Matthias Alexy states. He then elaborates on further model results. At the downstream edge of the dyke relocation area, a maximum rise in water level of 0.06m resulted in the vicinity of El-Km 483 as compared to the situation prior to the relocated flood dyke. This local water level increase is caused by the fact that around 85% of the water discharged over the dyke relocation area, amounting to 1385 m3/s, reaches the main river via the relatively narrow apertures 5 and 6. In comparison to the situation without dike relocation, however, significantly less water is discharged in the main river channel (-1170 m3/s), so that lower flow speeds prevail.
The water flowing in from the dyke relocation area through apertures 5 and 6 brings the discharge back from 2635 m3/s to 4020 m3/s over a short distance. Because of the low flow speed, the additional volume of water can initially only be discharged by means of an enlargement of the flow cross-section and thus a rise in the water levels. The difference between this higher water level and the water level further downstream results in a locally steeper water level gradient and higher flow velocities. Due to the increased water level in the main stream, the water level in the lower part of the dyke relocation area is increased to 0.08m above the water level in the main channel.
In addition to the computations for the vegetal cover in the year 2009, forecasts were also generated for various vegetation scenarios. E.g. in case of a hundred year flood and assuming fully developed vegetation in 2009 the model predicted a maximum water level increase in the upstream region of the dyke relocation area of around 0.12m compared to the [actual] vegetation state in 2009. However, the water level is still about 0.25m below that without the dyke relocation.
The water level situation at the Mödlich river island located downstream (El-km 485 - 487) remains unaffected by the dike relocation. The vegetation envisaged here (tree galleries on natural deposits of fines along the banks, scattered copses on the floodplain) exhibit a very low flow resistance, such that the water level rise - just a few millimetres - is negligible.
Around the Lütkenwisch river island that lies upstream (El-km 474 - 476), a significant decrease in the flood levels of around 0.3m at a discharge rate of 4020 m3/s due to the dike relocation is anticipated, such that the calculated rise in water level of about 0.01m owing to the planting schemes is practically negligible.
Flood: the practical test for the computational model
A flood in March 2010 enabled the first field measurements to be used for model validation. These took place after the dike relocation, which was completed at the end of 2009. 'Apart from a few discrepancies of a technical nature at the upper boundary of the model, we found a very good accordance between calculation and measurement', as Matthias Alexy reported. Subsequent flooding events in October 2010 (fig. 3) and in January 2011 confirm the good agreement between water levels and flow conditions computed in the model and those observed in the natural environment. As Alexy added: 'Our experience with the model is now flowing into the interdisciplinary research project KLIMZUG NORD. This will help us - together with our research partners - to develop a climate-adapted, integrated floodplain management system in the Biosphere Reserve Elbe Floodplain in the Lower Saxony Elbe Valley.' Based on the numerical model generated by the BAW, the impact of changed discharge and vegetation conditions on the Elbe´s water level are to be simulated using various climate change scenarios. 'We are currently basing our modelling on discharge rates of up to 5000 m3/s for future climate-induced extreme flood events.'