Hydrodynamic Modeling of the Lower Mekong River and Delta

For the past three years Halcrow Group Limited has been involved in developing a large hydrodynamic model of the Lower Mekong River including the Great Lake in Cambodia and the delta in Vietnam. The model has been built as part of a decision support system intended for use by the four member states of the Lower Mekong Basin: Thailand, Vietnam, Cambodia and Laos PDR. The system has been developed as part of the Water Utilization Program (WUP), which is intended to help the Mekong River Commission (MRC) member states to implement key elements of the 1995 Mekong Agreement. It will provide the technical and institutional capacities required for longer-term co-operation for sustainable management of the basin’s water and ecological resources.

Background

The complete project encompasses the data storage and handling for three complementary models covering the hydrological and hydraulic processes of the whole Lower Mekong Basin. These in turn are integrated within a user-oriented environment for testing and assessing impacts of developments and interventions in terms of flooding, water shortage and environmental impacts.

The Basin Modeling and Knowledge Base Development Sub-Project is being undertaken by Halcrow for the MRC. An important outcome of this project is a Decision Support Framework (DSF), which contains a Knowledge Base, a Basin Modeling Package and Impact Analysis Tools. Now completed, the WUP DSF is to be used to assist in developing rules for water sharing amongst the four riparian countries in the MRC and to support decision making for basin planning and management through assessment of the environmental and socio-economic impacts of development options.

Distinguishing Features of the Mekong Delta

The Mekong Delta has a number of unique distinguishing features that create a challenging task for the development of a comprehensive hydrodynamic model:

• A large proportion of the delta is liable to inundation for long periods of time affecting a significant proportion of the populated land area of Cambodia and South Vietnam.

• High out of bank flows occur such that peak flood flows in the mainstream are reduced downstream (Figure 1).

• Significant flows in the Mekong and the total inflow from the Tonle Sap Basin in Cambodia are stored within the Great Lake at the height of the monsoon period resulting in a reversal of flow of the Tonle Sap River at Phnom Penh twice yearly.

• Tides of around 2.5m amplitude at the mouths of the Mekong interact during floods and low flow with the very low lying land within Vietnam.

• Numerous major and minor canals distribute water to one of the most productive agricultural and fisheries areas of the world.

• Gates have been installed to control saline intrusion.

• Low embankments and major roads affect distribution of flood flows.

• Floodwater runs through and is stored on the higher ground of the Cambodian floodplain but at a critical stage in a large flood it rapidly floods across the border into Vietnam through a number of paths causing widespread overtopping of local banks protecting land from flooding (Figure 2).

Figure 1 - Attenuation and Alternative Flood Routes Change the Flow Significantly in the Mekong Mainstream (448km, 346km, 235km and 127km from sea)

Figure 2 - Satellite Imagery Showing Onset of Flooding 20 July 2000 and Progress of Flood 10 Days Later

Requirements for the Model

The hydrodynamic model is not required solely for use by experts for a flood study as might be a typical use of such a model. The tool is developed and maintained at the MRC but in use at a number of sites in the four riparian countries for:

• Simulation of flood regimes - for example, assessing changes due to dam construction upstream for which typical years or long term simulations may be carried out and compared with the base case

• Simulation of changes in saline intrusion

• Providing outputs for impact assessment tools including making use of GIS for spatial analyses

For practical use there is a limit on the times taken for simulations, which places a limit on the spatial complexity of the model that can be used, balanced against the need to represent the system adequately for planning purposes. Extensive discussions and tests on levels of schematization detail were carried out and agreement with the parties was obtained on a relatively complex model with around 5000 nodes comprising major rivers, canals and flood storage and conveyance. The node location diagram for the part of the network downstream of Phnom Penh is shown in Figure 3, which illustrates the complexity of the model.

Figure 3 - Example Schematization

Data Inputs and Model Calibration

In a hydrodynamic model the bulk of the data used relates to physical data such as river sections, banks or fields. The ISIS model was built up using GIS processing to extract data from digital models of the terrain and bathymetry derived from existing data sets within the MRC. Some data were also extracted from the existing VRSAP model used in Vietnam. The geographical location of the model nodes was integrated within the model, thus facilitating future updating of the model and enabling the model outputs to directly link into spatial tools for analysis of results. For example, the soundings of river depth from the hydrographic atlas of the major rivers in Vietnam and Cambodia were combined with land surveys of banks and ground levels to create a digital terrain model within the ESRI ArcView environment. River sections, embankments and flood storage units could then be abstracted directly into the model input data. Where river sections were not available in such a form, for example for smaller canals, the centre of each section was registered to the same coordinate system enabling direct comparison with satellite data on flood extent for example.

The model calibration stage covered both flooding and low flows for 2000 as a calibration year and 2001 and 1998 as verification events. The model was calibrated for 31 permanent water level stations on the mainstream and floodplain, and 5 permanent flow gauging stations. The calibration was assessed using threshold parameters indicating the goodness of fit to the field data for wet and dry seasons and year round statistics for flood peaks, mean absolute differences and tidal ranges. The approach taken is more rigorous than originally envisaged but provides an easier way to assess results than can be seen from comparison plots and quickly highlights areas for potential improvement. For the tide-affected stations the comparison of observed and model outputs was calculated using hourly intervals for the whole year. Sample time series comparison plots are shown in Figures 4 and 5.

Initially the model was calibrated separately for wet season and dry season periods, however, as the planning model had to run continuously for 16 years, it was necessary to calibrate for both dry season and wet season conditions within the single model. Depth variable roughness was introduced into the ISIS software and much effort was taken to produce a robust model that would run stably for long-term simulations with ranges of boundary conditions.

Figure 4 - Simulated and Observed 2000 Water Levels at Kratie

Figure 5 - Simulated and Observed 2000 Water Levels at Tan Chau

Salinity Modeling

The ISIS modeling package is also able to take the hydrodynamic results obtained with the full model and use these to drive a water quality simulation. For the Mekong, the most important water quality parameter is salinity, as saline intrusion into the river and canal system in the delta affects the use of water for irrigation. The preparation and calibration of the salinity model is relatively straightforward comprising primarily the definition of salinity boundaries and dispersion coefficients to be used. The SMART algorithm used in the ISIS Quality computation enables fast and accurate simulation of the advection-dispersion process. The main difficulties for development of a calibrated salinity model lay in the sparseness of data for definition of the boundaries and points for comparison. Within the DSF the results can be spatially processed, again in a GIS environment to produce, for example, contours of maximum salinity in a particular month as shown in Figure 6.

Figure 6 - Modeled Salinity Contours

Impact Analysis Tools

There is a range of environmental and socio-economic impact analysis tools that can be used within the context of the DSF. The foundation of the needs for Impact Analysis Tools lies with the transboundary issues of concern identified by the riparian countries. The main categories of transboundary issues are:

• Water Quality Deterioration and Sedimentation

• Fisheries Productivity and Ecosystem Functioning

• River Bank Erosion

• Obstruction to Navigation

• Inadequate Dry Season Flows

• Flooding

The analytical tools for environmental impact assessment required for the DSF need to be issue-specific and predictive in a manner that is replicable, objective, quantitative and sensitive to impacts. Quantitative tools are not confined to those that produce numerical results, but include those that rank, or categorize results based on analysis of numeric data. Expert knowledge, including traditional knowledge, is an integral part of many of these tools. Based on the identified issues and sub-issues, a range of generic tools has been identified for analysis of time-series data, spatial data and simple functional relationships of environmental responses. The principal limitation on the application of such first order impact tools is the level of current quantifiable knowledge of environmental responses to various land use, geomorphic, fisheries management, river flow and water quality changes.

The impact analysis tools developed for the DSF rely on the results of the hydrologic and hydraulic simulation models. For example, Figure 7 shows the impact of a hypothetical embankment on the Mekong left bank on flood durations (calculated by the ISIS model) - these data can be intersected with ecological or socio-economic data to assess positive and negative impacts.

Figure 7 - Impact of Hypothetical Embankment on Flood Durations

Conclusions

The Decision Support Framework and hydrodynamic model have been successfully set up in the Mekong River Commission Secretariat offices in Phnom Penh and in local offices in the four riparian countries. The ISIS hydrodynamic model successfully runs for the target 16-year simulations despite its complexity, and produces a representation of the system that is adequate and acceptable to the client. This has been a complex undertaking with the model development taking place within a consultative framework with representatives of the four countries each of which have their own views and requirements.

Author: Dr Jon Wicks, Halcrow Group Limited
Originally presented at the 2003 Wallingford Software International Users’ Conference