RiverWeb^SM Water Quality Simulator

The Computational Model

At the heart of the Water Quality Simulator is a simple computational model which calculates conditions in an archetypal watershed over time. The implementation is in Fortran, and the way the model interfaces with the simulator itself is covered in detail in the section written on Scientific Modeling in the Simulator. For reference purposes, it may be useful to refer to the glossary.

The model can be divided into the following five sections:

I. Initial Conditions
II. Hydrology (Runoff, Groundwater, and Evaportranspiration)
III. Nutrient Loading and Concentration
IV. Physical Indicator Calculation
V. Biology

I. Initial Conditions

Virtually any computational model needs initial conditions to run. In our case, a great deal of information is needed, some of which can be entered on the fly via the web interface [although this needs to be done prudently, because allowing more flexibility in setting initial conditions increases interface complexity]. We have data which has been collected on daily precipitation and air temperature at a site in Maryland over a number of years. In general, this type of data is available for many places in the United States on the web. (see, e.g. The National Climatic Data Center). For the purposes of our simulation, however, it is useful to examine how the model works with any given set of reasonable initial conditions, so the localization or actualization of initial conditions is not of utmost concern.

An important initial condition in our model, at least when examining the mouth of our "watershed", is the division of the overall watershed acreage into different land uses. Currently, these values are somewhat arbitrarily fixed. In the future, this may be an option that can be altered through the web interface. This option would allow a teacher or advanced student to localize the model to some extent, or demonstrate how altering land use distributions affect the net conditions at the mouth of a watershed. This would demonstrate how different growth patterns such as urbanization cause changes in water quality within our simulator.

II. Hydrology

A quick summary of the hydrological cycle may be useful: In the environment, precipitation which falls on the ground ends up either immediately running off over the surface (as "runoff"), or infilitrating into the ground. The flow that we see in streams in a normal dry period is what is known as baseflow, and is caused by groundwater entering the stream. This groundwater input into a stream is generally fairly constant although it may increase some after precipitation events, and the amount of input into a stream depends a great deal on the local geological conditions. Finally, water can be transpired by vegetation back into the atmosphere, or directly evaporate from the ground.

In our model, we use empirical models of the components of the hydrological cycle. To calculate runoff, we use the "curve number method" which bases its predictions of precipitation runoff on land use. [Simply, the more impermeable the surface is associated with a land use, the more runoff is caused]. Runoff instanteously enters the stream. Then, the balance of the precipitation that doesn't run off infiltrates into groundwater. Groundwater can evapotranspire, and does so at a rate dependent mostly on the land use being examined and air temperature. (The empirical method is adapted from the "Blaney-Criddle method"). Finally, the groundwater flow into the surface stream is roughly a time-averaged percentage of the total volume of groundwater. Click here for more details.

III. Nutrient Loads and Concentrations

As part of nationwide initiatives in the 1970s and 1980s to reduce non-point source water pollution, a great deal of study went into determining how land-use and nutrient production was correlated. One result of these studies was to cast light on how complex the impacts that humans have on natural watersheds are, because how we affect the water quality when we use the land was not always consistent. However, these studies did produce a great deal of "average" and "typical" effects of changing land use on water qualities. These data are called "event mean concentrations" (EMC's) for a given nutrient and land use. In general, these data are the source of the nutrient chemistry in our model. This data does not allow for a process-by-process assessment of nutrient chemistry, but instead represents an average of how the combination of natural processes in a normal stream affects average chemistry.

Using the EMC values calculated in these studies for a given land use, we could calculate how much of a given nutrient would be in our stream. For each land use, a different pair of EMC values for runoff and groundwater were used, because nutrients propogate differently in groundwater and runoff. Sometimes adequete EMC data were not available for a land use or nutrient. In these cases, we made estimates based on what we considered to be the dominant processes in the natural system. This section of the model will likely be refined with further research.

IV. Physical Indicator Calculations

One of the most easily examined properties of a stream in the field is its physical properties, such as average depth, width, velocity. These are the components of total flow or discharge, since (avg.depth)*(avg.velocity)*(avg.width)=Discharge. We know the discharge in the system from our hydrology modelling, which calculates daily stream flow. At present, we calculate these physical properties using regression-based power laws from Leopold's "A View of the River". The actual form of the power laws varies a great deal regionally and even locally in nature, so for our archetypal watershed, we feel justified in picking a precise form for demonstration purposes. As of yet, however, we have not altered the interface to allow for display of these characteristics.

Other physical characteristics such as water temperature are currently calculated, using a strictly back-of-the-envelope empirical model. The accuracy of this has not been ascertained, and further research will likely be needed to determine whether we are calculating water temperature reasonably. [This is a reasonably difficult problem, since in real life local conditions such as local slopes and local velocities can affect things, as can things as esoteric and difficult to model as wind over the water surface.]

V. Biology

We've made a first-order attempt to examine how water quality and land use affect biology within a watershed. We are currently using a rough regression-analysis based model for biology which is not truly causal. Because we have not yet added this information to the interface, and a great deal more development needs to be done on the details of our model of this difficult topic, what can be said usefully at this time is limited. More information will be added to this section at a later date.