Prof. D. Stephenson and Mr. M. Furumele
University of the Witwatersrand, Johannesburg
P O Box 277, WITS, 2050, South Africa
Tel: +27 11 717-7154, fax: +27 11 403-2062,
E-mail: steph@civil.wits.ac.za
Abstract: Increasing urbanization, in particular low-cost housing along the banks of streams, is leading to increased flooding. Consequent damage and health hazards need careful consideration. This planning process should be formalized in order to decide the localities for housing and other urban development.
The standard method of deciding water levels is to estimate floods from either an existing flow record or preferably by synthesising from rainfall records. The latter is preferable because the model can then account for increasing urbanization and other structural changes such as channelization, bridges, waterways and management structures.
The hydrologic analysis yields the flow rates with different probabilities of exceedance and a subsequent hydraulic analysis will indicate the corresponding water levels for different recurrence intervals. The latter is also dependent on the roughness of the channels and the geometric cross-section.
By coupling the hydraulic model to a GIS system, it is possible to depict the extent of the flood plains for various recurrence intervals with different colours. The risk of flooding of any locality is at plotted.
The hazard of flooding is another factor to consider. That is, whether there is avoidable damage, tolerable damage, serious damage or catastrophic damage. Thus an index of different hazard levels can be obtained. The hazard index should be coupled with the risk index in order to establish a hazard-risk index which in turn should indicate the desirability of development at different distances from waterway.
Keywords: floods, hazard, risk, rivers, waterways
Urbanization is becoming a fact of life and problem for many population groups, particularly in developing countries. The level of migration to urban and economic centres in many cases outstrips the rate of earning and servicing. The result is that there is great pressure for land, in particular for low cost housing and in many cases, informal housing clusters developed faster than planning procedures can advance. There are often catastrophes when floods occur or due to heavy pollution of the urban waterways. Therefore low cost management systems are desirable. However, more importantly is the demarcation of hazardous and risky areas at an early stage. The hazard is related to the degree of danger and the risk is associated with the probability of the hazard occurring. Both factors must be considered and the hazard-risk index is in fact what is proposed for plotting on development maps.
Urban floods are characterized by the changing nature of the catchment and waterways, as opposed to rural floods which are generally for natural catchments. It is therefore not easy to extrapolate existing gaugings even if they are available for urban streams because the characteristics, in particular the intensity of the flood and the management of the flows may change over time. Indeed, flood management by way of detention or retention is more important for urban streams as the floods increase due to urbanization.
Not only is there an increase in impermeable cover in the case of urban development, but also the impermeable cover such as roofs and roads is in many cases directly connected to the drainage network. The drainage network increases flow velocities due to lower friction factors and deeper channels. The consequence is that the critical storm duration is shortened compared with natural catchment and therefore a more intense design storm becomes the norm. This means that the spectrum of floods changes due to urbanization.
It is however often easier to install management structures in urban streams as the streams are smaller, but more particularly the economics dictates that management systems should be installed.
Therefore it is necessary to conduct rainfall-runoff modelling (e.g. Stephenson and Palling, 1992) when studying urban water systems in order to optimize the level of management and in order to calculate the water levels associated with different frequencies or occurrence or exceedance.
The analysis of rainfall records often yields long-term records statistics (e.g. Chang, 1989) and more particularly rainfall intensities for different duration storms which can be built into a model for rainfall-runoff calculations such as XP SWMM (1993) or RAFLER (Stephenson and Paling, 1990).
The design process from hereon is iterative as the level of management such as by detention or diversion storage influences the extent of flooding downstream as well as the backwater of the reservoir. In socio-economics optimum level of management for a selected time horizon is necessary. It should also be borne in mind that entire analysis should be done for future level of urbanization when flood intensities and volumes are higher.
The next stage in the method of analysis is to calculate flood levels using a hydraulic-type model such as RIVERCAD (Boss, 1977) or BACWAT (W.S.R.G., 1990). This requires contour information which can be obtained by digitizing contour maps or aerial photography. The resulting computed water levels can be replotted on digitized maps on a GIS system.
The resulting maps and cross-sections through the channels will indicate different water levels with different recurrence or risk of exceedance. That is, the lowest waterway is plotted usually that associated with a 1-year flood (Stephenson et al., 1997), whereas the channel thalweg is that associated with the average minimum dry season flow. Successively higher levels will cover the flood way and the flood fringe. The waterway is usually taken to be that within the bounds of the 1-year flood, whereas the flood way is taken to be that within the 20-year flood line and development is seldom prohibited below this level. The next higher level such as the 100-year flood could embrace the flood fringe and again only certain development would be permitted within this fringe subject to the sensible design and an authoritative consideration of the effects of construction within the waterway or flood fringe. I.e. the more development that occurs with these fringes, the greater the backup effect is, and therefore the wider the waterway and flood fringe become. It is often not easy to anticipate the type of development which will occur within these lines.
The different probability or risk zones are demarcated with a risk index. That is, above the 100-year flood line will be given a zero rating, between the 20-year and 100-year flood line will have a rating of 1, and below the 20-year flood line, the risk index will have a value of 2.
The level of flooding which can be tolerated depends on the socio-economic impact of the flooding. Thus there may be no hazard in rural areas if there is no people danger within the flooded area. The danger generally increases with deeper flows and faster flow velocities. Thus, the ideas developed in Minnesota (1969) were formalized by the New South Wales authority (1986) who indicated values of velocity x depth as giving an index.
Thus a velocity of 1m/s and a depth of 1m would result in a high hazard as buildings could be washed away. A value of the product of 0.5 could still be dangerous to people, particularly children. An even lower water depth, that is of the order of 200mm, could cause severe damage within buildings, but may not be a danger to life. This level may also be manageable provided a warning system exists and people can use sandbags or other diversion means to avoid extensive economic damage. A water level as low as 0.05mm would be a danger to traffic on roads.
So we have a range of hazards which are given the values of 1 for depths of flow below 0.2m, 2 for a higher depth but with a velocity x depth less than 0.5 and a value of 3 for greater than that.
The hazard-risk index is then obtained by multiplying the risk index by the hazard index. Thus a value of 0 to 6 is possible, where 0 is of little concern and 6 is of highest concern.
The authorities would therefore obtain from the GIS system the hazard-risk index before deciding whether to allocate land to different types of development. Thus hazard-risk indexes above 3 generally require constructed waterways and/or barriers erected along the fringe.
For successively decreasing hazard-risk index, different levels of development and in particular raised development may be tolerated but not encouraged. Careful monitoring of the fringes is therefore required (e.g. Semenya Furumele, 2000).
BOSS International (1977). RiverCad Modelling System Users Manual.
Chang, T.J. (1989). Characteristics of extreme precipitation. Water Resources Bulletin, 25 (5), 1037-40.
Minnesota (1969). Floodplain Management Act, Minneapolis.
New South Wales (1986). Flood Development Manual, Sydney.
Semenya Furumele Consulting (2000). Blesbokspruit Regional Development Plan, Johannesburg.
Stephenson, D. and Paling, W.A.J. (1992). An hydraulic based model for simulating monthly runoff and erosion. Water SA, 18 (1), 43-52.
Stephenson, D., Van Zantwijk, A. and Hoge, P. (1997). A flood plain development policy for the Vaal Barrage. Proc. Instn. Civil Engrs., Mun. Engr., London, 121, Dec, 199-205.
Water Systems Research Group (1990). Bacwat users manual, Univ. of the Witwatersrand, Johannesburg.
XP SWMM (1993). Stormwater management model Users manual, Florida.
Fig. 1 Flood risk diagram
Fig. 3 Channel zones
Fig. 4 Sample hazard risk plot