Riverweb: Simulator: Scientific Modeling

Simulator Functionality

Learner Interface

The RiverWeb Water Quality Simulator (WQS) depicts the effects of various land uses on water quality in an archetypal watershed. By limiting each sub-watershed to one land use, the effect of that land use can be seen on the quality of the water that students "test" within its boundaries. The cumulative effect of the combined land use determines the water quality shown by the indicator values found at the common outflow. After the user logs in, a map of the archetypal watershed appears.

Water quality monitoring stations located throughout the watershed are depicted in the map. The user may click on the map to investigate any sub-watershed. The RiverWeb graph window also appears. By default it displays the variation of nitrogen over time in the top window, and precipitation over time in the bottom window. As shown below, other indicators may be selected, or the user might compare nitrogen concentration between two stations.

In addition, reducing the day range for each graph provides the ability to zoom in on a particular time period. A scatter plot is available to further delineate the relationship of the pair of indicators or stations. A digital notebook link is keyed to questions related to the indicators currently selected. The Tour option, which may be selected at login, uses frames to combine the WQS with instructions leading the user through most of the simulator capabilities.

Clicking on a hyperlink in the graphical display evokes a web-based notebook which is linked to a flexible database on the server side. The notebook provides a space for students to record their observations, pose hypotheses, and answer questions designed to promote problem-solving as they explore connections between watershed variables. Teachers can use the notebook to structure their students' explorations by customizing the questions to fit the needs of their students and curriculum, and to assess student learning.

Scientific and Computational Model

The model behind the WQS was initially implemented using STELLA as part of an MVHS Core Models activity packet (Shaffer et al., 1998). Inputs to the model include yearly time series of precipitation and air temperature observations. The program applies research-derived land use curves provided by the USDA Natural Resources Conservation Service to the precipitation data to produce average runoff. Based on empirical "mean concentration" values measured for different land uses, the model calculates the loads and concentrations of the nutrients, sediment and toxins carried in the surface flow. The most complex causal relationship involves dissolved oxygen, which depends on water temperature and nutrient load, which in turn depend on other factors. During summer, 2000, the model equations were developed further in consultation with several scientists including Glenn E. Moglen, Department of Engineering, University of Maryland, College Park and Alexey Voinov, University of Maryland Center for Environmental Science. For example, groundwater calculations were revised to include lag relationships, and evaporation and soil variation were added to the model. Development of the model is ongoing, as a middle course is found between a teaching model and a hydrological model. A more complete description of the model is available.

Equations from a revised STELLA model were used as the basis of source code written in Fortran and compiled on the MVHS web server. The RiverWeb frontend, rewritten in PHP in the summer of 2005, takes the data file generated by the Fortran executable and creates graphs and a user interface to control the data plot. Indicators and stations are selected by the user. The server-side executables provide not only time series graphs showing change in a selected indicator over time, but also scatter plots depicting quantitative relationships between two indicators, for example sediment versus runoff. The scatter plots require that the selected time range in days be the same for both indicators. However, scatter plots of lag relationships may be created for one time series when offset from the other.

The server-side capabilities described above support calls from multiple clients simultaneously, since all files are named by the process ID of the script invoked when the user submits a graphical display request to the server.

Last Modified: January 2006