Contamination of Groundwater-Fed Springs in Uganda
by Leigh Soutter
Robinah Kulabako (left) negotiates about her work in a contaminated spring with an inhabitant of the peri-urban settlement Bwaise III near Kampala (Uganda). To the right a boy carries containers full of the spring water.
Water contamination within the densely populated lowlands fringing Kampala (Uganda) threatens the health and livelihood of uncounted thousands of inhabitants. With no formal infrastructure of almost any kind, residents of these reclaimed wetlands drink contam-inated water from open springs. Dr. Roger Thunvik and Robinah Kulabako of the Royal Institute of Technology, or KTH, in Stockholm (Sweden) use COMSOL Multiphysics to define just how the contamination evolves so they can stamp out the problems with low cost remediation schemes.
Just around the edges of Kampala (Uganda) sit a number of informal settlements, densely populated low income communities that exist more or less off-grid. Inhabitants may work amid the relative urban wealth but live mostly without any of its formal infrastructure for water, sanitation, energy. These peri-urban developments account for 60% of the city's residents.
Situated in the lowlands, the water table in peri-urban Kampala lies only a meter or two below the ground surface. Nevertheless dwellers here use on-site septic tanks and pit latrines. They dump wastes directly on the ground or in unlined pits and gutters as convenient. These contaminant sources often sit directly upgradient from supposedly protected ground-water fed springs.
The worst contamination troubles come during the rainy season, when infiltrating water carries the wastes directly to the groundwater supply. While the shallow water teams with fecal matter more than 90% of the peri-urban poor take all of their water from springs the groundwater feeds (Figure 2). If they treat the water at all, they do it in their homes. People get sick, but no government authority monitors any of this.
Figure 2. A protected spring in Bwaise III, a peri-urban development near Kampala (Uganda).
Bwaise III sits within the peri-urban zone about 4 km north of Kampala city center. It fringes on swampland to the west of Lake Victoria. With 27,000 people per square kilometer, it has one of the highest population densities peri-urban Kampala. Typical of the peri-urban development, it has a very shallow water table (< 1.5 m), and serious problems with shallow groundwater contamination.
Zeroing in on the contamination
Dr. Roger Thunvik, professor at the Royal Institute of Technology in Stockholm (Sweden) and Ms. Robinah Kulabako, a lecturer at Makerere University in Kampala (Uganda), have been studying the shallow groundwater contamination for about two years. In addition to collecting high quality data with a thorough field campaign at Bwaise III (Figure 3), they recently started modeling the shallow groundwater contamination problems with COMSOL. They use COMSOL because they need to back up the conceptual model built from their measurements with a quantitative description of the physical processes that cause it. The flexible format allows them to change their ideas and assumptions readily, without building new models.
Figure 3. Water quality data from a protected spring in Bwaise III. Waste contaminants coliform (red) and phosphorus (green) vary with rainfall. Otherwise discharge to the spring (black) and natural salt levels (brown) are more or less constant.
Quoting Dr. Thunvik, "Experimental observations are just results from physical processes, like a few pieces of a jigsaw puzzle fortuitously positioned in the right places. Since these measurements don't even give you a complete snapshot of the contamination or even explain its causes and effects, you can draw the wrong conclusions from experimental observations alone. With COMSOL Multiphysics, however, we explain how rainfall rates, water saturation, the water table depth, soil properties, and starting concentrations interact and evolve into phosphorus plumes. With the right explanation it's straightforward to devise strategies to remedy the problem and put them into action. Teaming up with Dr. Mai Nalubega from the World Bank, we see our efforts going to a good end."
The Bwaise III study area
If Bwaise III were in Florida you'd probably call it a swamp. Situated in lowlands, the groundwater table mostly lies about one-half meter below ground but outcrops at the surface in places. There the water pours out of the ground at natural lows and in ditches, especially during the rainy seasons from March to May and from September to November (Figure 3). They get between 1.1 m and 1.4 m of rainfall per year.
Locals reclaimed the Bwaise wetlands and settled it by adding man-made loam over the swampy ground (Figure 4). At Bwaise, the shallow groundwater that feeds the streams circulates in silty clay layer just below the man-made loam. The silty clay aquifer overlies a relatively impermeable layer of stiff dark clay. The bedrock, about 30 m below the clay, contains clean water in a fracture network. But being so deep and haphazardly distributed in hard rock, the locals cannot access it without machines.
Figure 4. Profile showing the impermeable clay bottom (Layer 1), the shallow groundwater aquifer (Layer 2), and the man-made loam (Layer 3). Also shown are the conditions on the cut out section from the larger COMSOL Multiphysics model.
By installing observation wells in shallow groundwater aquifer Ms. Kulabako measured how the water table depth and the concentrations of the different contaminants vary in the over a 19 month period (Figure 3). During that time, she also sampled contaminant concentrations directly from the spring. With the pit latrines, the solid waste and sullage dumps, the animal yards, and the car washing bays and garages, the contaminants mostly proved to be non-industrial waste. Given the alarmingly high concentrations of coliform at 1-16 x 107 cfu/100ml, nitrate at 0.10-779 mg/l, and phosphorus at 0.001-13 mg/l, the major pollutant sources appear to be the pit latrines. The highest concentrations appeared after heavy rains as expected - so the fecal waste infiltrates with the rain and leaches through the unsaturated zone to the groundwater.
About the data collection Ms. Kulabako explains, "A major and unexpected hurdle was suspicion from the locals. The Bwaise III inhabitants didn't understand what we were doing. Once we learned to communicate to them calmly, with very basic language, our efforts sped up."
Modeling the contamination of groundwater-fed springs
Dr. Thunvik and Ms. Kulabako currently focus on describing how the phosphorus moves through the unsaturated soils to the aquifer below and eventually feeds the springs. They started with phosphorus because it is a good proxy for the rain-driven contamination problem - it responds to the precipitation rates and derives solely from the local waste.
Their models couple the Richard's Equation analysis for variably saturated fluid flow and the Solute Transport analysis for the phosphorus transport, both from the Earth Science Module of COMSOL. They focus on a vertical cross-section 4.2 m long and 1.45 m deep cutting through an area that slopes at 1.5% to a protected spring (Figure 4). Unclear about all of the details on the boundary conditions, they situated the focused section as a cut from a larger model covering a 12 m length. Then they fine tuned the boundary conditions on the larger model until it matched the monitoring well data and the spring concentrations at the focused cut out.
Unsaturated-saturated flow
In modeling the subsurface flow an important issue was how to handle the varying rainfall boundary. Initially Dr. Thunvik and Ms. Kulabako ran a steady state variably saturated flow simulation to start the model and added a constant rainfall rate to describe what happens for a single outburst (Figure 5).
Figure 5. This animation shows the flow field that evolves when water infiltrates into dry soil for almost two days from 0 m< x <6 m: effective saturation (surface), water table (contour), velocity field (streamlines). The rain infiltrates at a constant rate into the exposed ground surface and trickles to the aquifer. The water table rises initially and reaches almost steady state after about 1.5 days. Click the figure to start the animation.
Dr. Thunvik explains, "Next we examine intermittent rainfall. Unlike any other software program I've used, COMSOL Multiphysics lets you switch between a flux condition like rain and a specified ponding height. A couple of the example models that came with the Earth Science module have this type of boundary set up. We'll apply those strategies when we model strings of outburst separated by dry spells."
Chemical transport and reaction
For the phosphorus transport, they include the Solute Transport application mode in their COMSOL model file. Specifying only the concentration of the incoming rainfall and the initial condition, they use the Advective transport boundary condition, and COMSOL Multiphysics finds the concentration at all of the flux boundaries. They simulate the phosphorous transport for the duration of the rainfall event, usually less than 1 day (Figure 6).
Figure 6. Animation showing how the phosphorus moves with the infiltrating water to the spring: concentration (surface), water table (contour), velocity field (streamlines). Even though the contaminant concentrations are constant at the ground surface, the aquifer (below the contour) receives higher phosphorus loads as the water table rises. Total simulation time is 2 days. Click the figure to start the animation.
In their first models, Dr. Thunvik and Ms. Kulabako assumed that the phosphorus simply flowed with the water. In reality the phosphorus also attaches to and detaches from soil particles and organic matter during transport (Figure 7). Their model is very similar to the Sorbing Solute example on this CD except they switch to the Langmuir and Freundlich isotherms, which the Solute Transport application mode automates.
Figure 7. Snapshot of the subsurface from one of the concept-development simulations with sorption: aqueous concentration (surface), solid concentrations (contours), velocity field (streamlines). The sorbing plume moves slower and at lower concentration than the plume without sorption because of the time spent attached to soils.
Estimating the coliform and nitrate pollution requires modeling chain reactions and metabolic process using strategies like the one set up in the Reaction Chain model included with this CD. They'll add applications for the coliform and nitrate chain to the same model file they work with now and press solve.
What's left
Dr. Thunvik and Ms. Kulabako follow a careful modeling strategy. They first develop a conceptual model for how typical water flow conditions in this area affect the spread of these contaminants in general. Next they modify the general analyses to characterize contamination during particular periods and validate the model by matching to data. They will use the validated models to develop strategies for remedying the current pollution problem and to protect the shallow groundwater from future trouble. This article describes some of their concept-development models.
In the future, Dr. Thunvik and Ms. Kulabako is validating their working explanation of the contamination problems by matching their COMSOL simulations to test data. After that, they use the validated model to develop effective strategies to remedy the current pollution problem Bwaise III and protect the shallow groundwater from future trouble.
Quoting Dr. Thunvik, "What really excites me about this project is that our modeling is going to a good end. Of course we will use our results to clean up the groundwater problems at Bwaise III. But since Bwaise III typifies the contamination within all the peri-urban settlements in general, we can use our results to remedy this serious contamination that skirts around all of Kampala and is making millions of inhabitants sick."
About Dr. Thunvik and Ms. Kulabako
Dr. Roger Thunvik's environmental research in Africa began about eight years ago. He is a professor in the Department of Land and Water Resources at the Royal Institute of Technology, or KTH, in Stockholm (Sweden). A long record in numerical modeling of environmental process, Dr. Thunvik started writing code to solve unsaturated flow and transport problems using the finite element method twenty years ago. Now he uses COMSOL instead.
Ms. Robinah Kulabako's work on water and wastewater quality includes designs for low cost water treatment, solid waste management. She has been an assistant lecturer with the Department of Civil Engineering Makerere University in Kampala (Uganda) for the past five years. She followed up her undergraduate work in Civil Engineering from Makerere University with a degree in Environmental Engineering from The University of Manchester (the United Kingdom). Now she is pursuing her PhD with Dr. Thunvik at KTH.