Managing Flood Risk On the Orange River, South Africa

March 19, 2014
A group of international water experts have teamed up to provide hydraulic modelling and flood management solutions for the Lower Orange River.
Boegoeberg dam

A group of international water experts have teamed up to provide hydraulic modelling and flood management solutions for the Lower Orange River. The project includes one of the largest hydraulic models in the country detailing over 350 km of river, covering the vineyards and wheat fields from Boegoeberg Dam, near Groblershoop, to Onseepkans.

By Stuart Dunsmore & David Ramsbottom

The Orange River is the largest river in South Africa. Its catchment area of almost 900,000 km2 reaches into Lesotho, Namibia and Botswana and yields over 20% of South Africa's fresh water. Rainfall varies dramatically from over 1000mm per annum in the eastern highlands in Lesotho to less than 100 mm per annum along the Atlantic seaboard. As a result, over 90% of the catchment's mean annual runoff is derived from the upper areas of the catchment, including the Vaal River which is a major tributary of the Orange River.

This yield has provided an important water source to an otherwise arid area along the lower reaches of the Orange River and since the late 1800s an important agricultural economy has developed along the river.

Flood control

The first of the weirs to be constructed was the Boegoeberg Dam, 150 km upstream of Upington. Plans for the dam and associated canals were first conceived in the 1870s but the full benefit of the scheme could not quite match its cost. This was reviewed on a number of occasions with varying areas of irrigated lands to be served.

It wasn't until the Great Depression in the late 1920s that the feasibility of the scheme was realised. Used as a job creation initiative, the 9 m high Boegoeberg Dam was completed in 1932 and irrigated 7,700 ha. Since then, another eight government irrigation schemes have been established downstream, with over 400 km of canals irrigating more than 20,000 ha.

Flooding would have been part of the agricultural life cycle on these lands and flood levees were a likely method of flood control. Early records in the Department of Water Affairs archives report on damage to flood levees after a flood event in 1957. However, it was only after a particularly large flood in 1974 that there was a concerted effort by the Department of Water Affairs and various local municipalities to develop a network of flood levees.

At that time, the design level for the levees was referenced to recorded flood levels during a smaller event in 1976. This did not take into account other flood conditions that could occur.

Subsequent events have shown that this has had variable success, with smaller floods resulting in higher flood levels in some locations. In addition, poor design and construction has resulted in slippage and failure of some of the levees.

After widespread damage during a flood in 2011, the Department of Agriculture, Land Reform and Rural Development (DALRRD) set out to address the problem. ILISO Consulting was appointed to undertake the hydraulic analysis, master planning and concept design of a large area of the irrigation system in the Lower Orange River, and to coordinate a programme of design and construction.

2D hydraulic modelling

The study area includes approximately 350 km of the Orange River from Boegoeberg Dam to Augrabies Falls, and a smaller 20 km section further downstream at Onseepkans on the border with Namibia. The DALRRD required a two dimensional (2D) hydraulic model to be developed for the study. ILISO Consulting appointed Fourth Element Consulting to undertake the 2D modelling work, while HR Wallingford was called upon as specialist advisors for both the modelling and flood management.

A 2D model was selected rather than the more familiar one-dimensional (1D) model. This was due to the complexity of the river system and the requirement to realistically model flood spreading across the floodplains caused by breaching and overtopping of defences at any location. Parts of the river system have a complex network of braided river channels with islands between the channels, some of which are irrigated and have levees.

The area to be modelled is just over 1000 km2 and a flood hydrograph in the Lower Orange system is typically a double peaked event of 30 to 45 days duration. A modelling strategy had to be developed to simulate the passing of a flood hydrograph along the length of the river in a sensible run time whilst maintaining sufficient detail in the model.

The InfoWorks ICM software has been selected for the study. The river has been divided into six models. The last one is a stand-alone model for the Onseepkans area, while the first five flow from one to the other from Boegoeberg to Augrabies Falls.

The largest model surface area is over 200 km2 and the detail is aimed at master planning rather than detailed design. This still requires that the flood levees are represented in sufficient detail to ensure that their effect on the passing flood hydrograph is adequately simulated.

Failed levees

One of the concerns of the DALRRD is that, in the past, the construction of levees has been overseen by different organisations including the farmers themselves, and the standards of design and construction have been variable. In many cases the levees were constructed from the sediment deposited on the lands in the most recent flood and are therefore prone to slippage and collapse.

Most of the levees are constructed right on the river bank (to maximise crop coverage), resulting in stability problems as well as being prone to erosion. In the 2011 flood it is understood that many of the levees failed before they were overtopped.

Data collected for the study included a LiDAR survey of the flood risk areas. There is no formal record of the flood levees so the LiDAR survey data has been used to identify them as far as possible. There are over 800 km of levees. Formally constructed and maintained levees are relatively easily identified from the LiDAR data, but this has not been the case everywhere and in some cases there were gaps in the flood levee data. The way in which these gaps are filled will depend on the impact that they have on the model results obtained during the calibration process.

While there will be uncertainties in terrain and flood levee conditions prior to the calibration events, the study will benefit from a comprehensive set of flood records for one of the largest flood events in recent memory. The flood of 1988 caused widespread damage not only in the Lower Orange River area, but also further upstream where there was widespread inundation, riverside towns flooded, croplands destroyed and even two dam failures.

The flood flow varied along the river depending on the impact of the floodplains, but was of the order of 8,000 m3/s in the modelled reaches. It is regarded as being close to a 50-year return period event. The Department of Water Affairs at the time commissioned the establishment of a series of flood beacons along both banks of the river and the data for these are being used in the calibration of the models.

Developing a master plaN The primary purpose of the calibrated models will be to support the development of a master plan for flood management along the river, with a specific focus on agricultural lands.

The master plan will also highlight vulnerable areas; those that are not currently protected by levees and those that could experience worsening flood risk and flood damage if upstream and surrounding levees are raised. The study is also starting to raise questions about how the design return period is going to be selected.

One of the challenges of the project will be to achieve an equitable solution for a long length of river where the levees and floodplain inundation have a significant impact on flood flows and water levels further downstream.

The question of drainage of the floodplains following inundation will also need to be addressed as this affects the duration of flooding and the impacts on crops. There are currently relatively few flood gates in the system. They typically consist of hinged flap gates but these are viewed as having variable success because of the possibility of failure during flood events. Manually operated sluice gates are also in operation at some points. Elsewhere farmers will break sections of the levees to release retained flood waters after the flood hydrograph has passed.

No doubt additional questions will arise as the master plan develops. It is expected that this will highlight the value of the asset that will be the 2D hydraulic model, not only for the present study but also the future management of the system. The model is possibly the largest of its kind in southern Africa and, as the benefits become apparent, may lead to similar modelling approaches for other major rivers in Africa.

Stuart Dunsmore is managing director, Fourth Element Consulting, South Africa and
David Ramsbottom is technical director, Floods, HR Wallingford, United Kingdom. Email: [email protected] and [email protected].