The purpose of this blog is to provide a guide to the groundwater infiltration module (GIM) within Infoworks ICM and legacy Infoworks CS. The blog will begin by describing why the GIM is important and focuses on the limitations of existing overland runoff models before introducing the equations used. A worked example is provided of how to run the model, how to view and understand the results and what effect changing the parameters will have.
The purpose of the GIM is to provide a highly attenuated response to rainfall, by representing the below ground processes of infiltration through the soil and contribution from a water table. It can be implemented in any type of catchment including storm, combined or foul / sanitary.
Figure 1 below shows a typical sewer inflow from a catchment incorporating the GIM. The descriptions relate to the InfoWorks definitions of these responses.
There is a lot of confusion regarding the terminology relating to a catchment response to rainfall. The following is a description of what the InfoWorks ICM components presented in Figure 1 above are intended to represent.
- Fixed or Wallingford. This is the fast, direct response from impermeable areas such as roads and roofs.
- New PR equation. This is the medium, direct response from permeable areas such as lawns and grass verges. These areas contribute overland to the nearest manhole.
- Soil store. This is a slow, indirect response through the soil and into cracks in the pipes.
- Groundwater store. This is a very slow, indirect response caused by water table rise as a result of rainfall.
- Baseflow. This is a constant flow, unaffected by rainfall.
In reality the distinctions between the categories are obviously blurred. In some countries, including the UK, the term ‘runoff’ is used to collectively describe the fast and medium direct responses. The term ‘groundwater infiltration’ is used to describe the slow, soil store and very slow groundwater store responses. The term ‘infiltration’ is used to describe the constant baseflow.
Before starting to understand the GIM, it is necessary to understand runoff. Runoff is the process whereby rainfall lands on the catchment surface and runs overland to the nearest manhole, gully or interceptor and from there into the sewer. This is described as the above ground response, whereas the GIM provides the below ground response. There are three basic components to runoff.
- Volume:- Runoff volume defines the proportion of the rainfall landing on the catchment surface which will enter the sewer. There are several different runoff volume models available in InfoWorks ICM. The simplest of these is the fixed runoff model, where a fixed percentage of the rainfall enters the sewer and the remaining percentage provides no contribution. In other runoff models such as the New PR equation or the SCS model, the proportion of runoff varies as a function of antecedent conditions.
- Routing:- Runoff routing defines the speed with which rainfall enters the sewer. Again there are several models to choose from in InfoWorks ICM including the Wallingford and the Desbordes models. They act like a unit hydrograph to attenuate the inflow.
- Initial Losses:- Initial losses represent the effect that the first few millimetres (fraction of an inch) of rainfall landing on the catchment surface is lost in processes such as damping and cooling the catchment surface and filling up small surface depressions.In InfoWorks ICM each of these three processes can be manipulated independently. For example, it is possible to change the speed of runoff without affecting the volume.The three processes together can be thought of as a reservoir with two outlets from it. Water starts to fill the reservoir until the outlets are reached. This represents the initial losses. Water passing through only one of the outlets goes to a defined location (into the sewer) whereas the other outlet leads nowhere. This defines the volume. The sizes of the outlets represent the routing. This concept is usually referred to as a reservoir model.Modellers frequently need to increase the ‘tail’ of the storm response in order to replicate a large observed flow occurring several hours after a storm has ended. Before the GIM was available the only method for doing this was to alter the parameters in the runoff equations outside the range for which they were originally intended. The New PR equation in particular was often manipulated in this way. However, the highly attenuated tail is often produced by below ground processes and therefore should not be replicated by calibration of runoff models intended for above ground processes.
GIM Model Overview
The most common use of the GIM is to provide a large, highly attenuated response to rainfall. Such a response is often observed in flow survey data. It originates from ground water, or rural or undeveloped parts of the catchment.
With runoff (described in the section above) there is some rainfall which enters the sewer and some which does not. The key to the GIM is that it allows a proportion of the rainfall which does not get into the sewer as runoff, to get in by other means.
The GIM provides two additional mechanisms for rainfall to get into the sewer. The first mechanism is used to represent the process whereby rainfall can seep through the soil and enter the sewer through cracks in the pipes. This is called the soil storage reservoir. The second mechanism is the process whereby the rainfall causes the water table level to rise. When the water table level exceeds the pipe invert level, there is a further contribution to the sewer. This is called the groundwater store reservoir.
A conceptual representation of the model including above ground runoff is shown in the Figure 2 below.
Both the soil store and groundwater storage reservoirs act in a similar way to runoff. They both have processes which closely resemble volume, routing and initial losses.
The following assumptions are made about the GIM.
- The reservoirs act independently. In other words the ground water level could be higher than the soil store level.
- The GIM takes no account of the flow depth in the pipe. There could therefore be infiltration inflow into a pipe that is surcharged.
- There is no interaction between ground water levels in subcatchments. The water table levels could be very different in adjacent catchments without any transfer of flow between them.
The Soil Store Reservoir
The parameters which affect the soil storage reservoir are as follows:
- Soil depth
- Porosity of soil
- Percolation threshold
- Percolation coefficient
- Percolation percent infiltrating
- Initial soil depth
- Rainfall evaporation
The soil storage reservoir can be thought of as a reservoir, filled with sand with one outfall from it. The height of the reservoir is the soil depth and the area is the same as the runoff area (i.e. the sum of the twelve surfaces, usually the contributing area.) The initial soil depth (defined in the groundwater infiltration event) and the percolation threshold are expressed as a percentage of the soil depth.When rainfall enters the soil store, the soil store depth rises. The rise is inversely proportional to the soil porosity. So for example if the porosity = 40%, an inflow of 1mm will produce a rise in the soil store of 1/0.4 = 2.5mm.
When the soil store depth exceeds the percolation threshold, there is a contribution from the soil store. The proportion of the flow which enters the sewer is defined by the percolation percent infiltrating. The remainder of the flow goes into the groundwater store. The rate of flow into the sewer is defined by the percolation coefficient according to the equation:
The soil store is also drained by evaporation. When the soil store is 100% full the evaporation rate is that defined in the rainfall event. When it is 0% full, the evaporation is zero. It is linearly interpolated for all values in between.
The Ground Store Reservoir
The parameters which affect the groundwater storage reservoir are as follows:
- Porosity of ground
- Baseflow threshold level and type
- Infiltration threshold level and type
- Baseflow coefficient
- Infiltration coefficient
- Initial groundwater level and type
The groundwater storage reservoir can be thought of as a reservoir filled with sand with two outlets from it. One outlet goes to the sewer and the other goes out of the system and provides no contribution. The infiltration threshold defines the level at which there is contribution to the sewer and the baseflow threshold defines the level at which flow is lost from the system. Either of these thresholds can be expressed as an absolute level or relative to the upstream invert of the pipe to which the subcatchment is attached. The infiltration threshold is generally higher than the baseflow threshold.
Input to the groundwater store comes from that which does not contribute to the sewer from the soil store. The starting point for the level in the groundwater store is defined as the initial groundwater level in the groundwater infiltration event. This can also be an absolute level, or relative to the invert level of the pipe.
The porosity of ground acts in the same way as the porosity of soil, a ground porosity of 40% means that a 1mm input from the soil store, will cause the ground store to rise by 2.5mm.
The rate of flow from the groundwater store into the sewer is defined by the infiltration coefficient according to the equation:
The rate of flow lost from the groundwater store and providing no infiltration into the sewer is defined by the baseflow coefficient according to the equation:
The groundwater store level increases and decreases as a function of the inflow from the soil store and the outflow to the sewer and to the ‘lost’ component. However it is possible to replace this process by manually defining the groundwater store level in the groundwater event.
Running a Simulation Including the GIM in InfoWorks ICM
The following shows how to implement a GIM response from a very simple one catchment model. Two elements are required, GIM catchment parameters and a ground infiltration event. Figure 5 below shows the catchment parameters.
Every subcatchment contains a Ground Infiltration ID. This links the subcatchment to the GIM parameters defined in the Ground Infiltration tab. Therefore each subcatchment could have different groundwater infiltration parameters. If no ground infiltration ID is defined then there is simply no ground water infiltration from that subcatchment.
Figure 6 shows a groundwater infiltration event.
A groundwater infiltration event must be included in a GIM simulation. It defines the initial level of the soil and groundwater storage reservoirs. From Figure 6 it can be seen that an initial soil saturation of 4% and an initial groundwater level of 40.4m have been defined in the sub-event parameters. These values will be applied to all catchments using the GIM unless superseded by parameters applied to individual subcatchment. Figure 7 below provides an example.
From Figure 7 it can be seen that subcatchment ABC would take an initial soil saturation of 5% and an initial groundwater level of 39m. All other subcatchments would use the values defined in the sub-event properties.
The groundwater infiltration event can also be used to define the groundwater store level, to represent a variation in water table level due to a river or sea. This replaces the groundwater store level calculated as a function of rainfall. In Figure 8 below a groundstore level is specified for catchment ABC whereas for all other subcatchment the groundstore level will be calculated as a function of rainfall with a starting level defined in the sub-event properties.
Understanding the Results
It is possible to graph the inflow to, the level in and the outflow from each of the reservoirs, as shown in Figure 9 below.
The sum of the soil store inflow and the ground store inflow is shown on the graphs as the groundwater inflow.
Below is a worked example which relates the defined input parameters to the response that they produce. Table 1a below presents the GIM catchment parameters. Table 1b presents the event parameters.
The runoff parameters are a fixed volume of 21.5%, zero initial losses and a routing coefficient of 0.001, leading the virtually instantaneous runoff. The catchment has an area of 10 hectares
Figure 10 below shows that the event used in this example has a depth of 36.16mm. With a catchment area of 10 hectares and no initial losses, this equates to 3616m3. The fixed 21.5% runoff means that 777.64m3 enters the sewer as direct overland runoff and the remaining 2839.29m3 enters the GIM as shown on the graph ‘infiltration to soil store’.
Infiltration to the soil store causes the level in the soil store to rise. The soil store depth is 1m and the initial soil saturation (defined in the groundwater infiltration event) is 4%. Therefore the starting soil depth is 4% of 1m = 0.04m. The percolation threshold is 5% and therefore it is only after the soil depth reaches 0.05m that there is a contribution from the soil store. The percolation percent infiltrating is 20% and therefore 20% of the flow goes into the sewer. The remaining 80% goes into the groundwater reservoir.
Figure 11 below shows that the soil store starts at 0.04m and reaches 0.05m after about 2.5 hours. It is at this time that the soil store inflow and infiltration to ground store starts. The infiltration to ground store is four times (80%/20%) greater than the soil store inflow. The sum of the infiltration to ground store and the soil store inflow is 2433.23 m3. This is less than the 2839.29m3 infiltration to soil store because some flow is lost before the soil store depth reaches the percolation threshold, in a process that can be thought of as initial losses. After the storm, the soil store depth reduces. It cannot fall below the percolation threshold (of 0.05m) because there is no evaporation defined in the rainfall event.
Infiltration to the ground store causes the level in the ground store to rise. The starting level of the ground store is 40.4m, which is the initial groundwater level defined in the groundwater infiltration event. Figure 12 below shows that it starts to rise at about 2.5 hours, when the infiltration to ground store starts. When the ground store level exceeds the infiltration threshold level, flow is lost to groundwater and provides no contribution to the sewer. In this example the baseflow threshold is also set to 40.40m, and therefore the lost to groundwater shown on the graph starts at 2.5 hours. The infiltration threshold level is set to 40.41m. When this level is exceeded there is a ground store inflow. From the Figure 12 below, it can also be seen that the ground store level exceeds 40.41m after about 4.5 hours and therefore the ground store inflow starts at this time.
The total of the lost to groundwater and the groundstore inflow is 1949.42m3. This closely matches the infiltration to groundstore. If the initial groundstore level is set lower, there will be a difference in the results again because of a process similar to initial losses. After the storm, the ground store level reduces until the baseflow threshold is reached. There is no way the ground store can fall below this level.
Figure 13 below shows the groundwater inflow (1523.46m3) which is the total contribution from the GIM. It is the sum of the soil store inflow (486.65m3) and the ground store inflow (1035.66m3). There is no dry weather flow in this example and therefore the total inflow to the node (2301.10m3) equals the runoff (777.64m3) plus the groundwater inflow (1523.46m3).
Figure 14 below shows the effect of manually specifying the groundwater store level. The defined groundstore level used in this example is the one shown in Figure 8 above.
Figure 14 above again shows that the ‘lost to groundwater’ starts when the groundwater store level exceeds 40.40m and the ‘ground store inflow’ starts when the level exceeds 40.41m.
Tables 2a and b below presents a summary of each parameter including advice on the effect of changing it.
The groundwater infiltration module is present within Infoworks ICM (and legacy Infoworks CS) and provides an approach to complement the runoff to achieve any required flow. The GIM is often avoided because it is considered too difficult to apply. However by following some fairly simple steps, a good level of calibration fit can be achieved.
- Produce a good a fit as possible by choosing the correct runoff model. The GIM cannot help in matching the early part of the storm response.
- Make frequent use of the graph tool to check the model response parameters shown in Figure 9.
- Apply the concepts of volume, routing and initial losses. To increase the volume, increase the percolation percent infiltrating. To increase the routing (to slow down the response) increase the percolation coefficient. To increase the initial losses (to delay the onset of contribution) increase the gap between the initial soil store depth and the percolation threshold.