A New Paradigm

IGW adopts a new paradigm to groundwater modeling. 

The basic concept  is very simple. But the impact is significant. Instead of treating flow and transport modeling as two separate steps, we perform them concurrently as these processes occur concurrently in nature. Instead of treating regional-scale modeling, local-scale modeling, and site-scale modeling as different phases in a long sequential process, we couple the multi-scaled processes on-line and model them simultaneously. Instead of treating visualization, analysis, and presentation as a “post mortem” analysis, we integrate them on-line dynamically and intelligently process the results on the fly. Specifically, we propose and adopt the following new modeling paradigm

For each discrete time step, at t = t_n

1.      flow modeling

2.      particle tracking, if contaminant particles are introduced

3.      transport modeling, if contaminant concentration plumes are introduced

4.      subscale flow and transport modeling, if detailed dynamics are needed in certain areas

5.      processing and mass balance and water budget analysis and other analyses

6.      spatial overlay, visualization and presentation

Repeat steps 1 to 6 for t = t_n + dt

The overall looping and model algorithm is programmed using an interactive visual language. The incremental flow and transport computations (for one time step) are programmed in Fortran and embedded in the overall program as a dynamic-linking library (DLL). Although the new software paradigm involves essentially the same modeling components, the same amount of numerical and graphical computations and processing, they are restructured and integrated into a single application program. All intermediate operations, often duplicated from project to project by countless modelers, are performed on-line in memory once for all. The result is a comprehensive interactive environment that makes it possible, for the first time, to perform real-time interactive groundwater modeling, real-time visualization, real-time analysis, and real-time presentation.

Real-time Visualization and Mapping.

Modeling based on the new paradigm eliminates the boundaries and disconnect among the different components, and create capabilities that provide a system view and management ability; it eliminates the time-consuming and error-prone intermediate processing, establishing real time communication between the assumptions/concepts and final results/implications.

The new modeling paradigm involves no off-line post-processing, no disk writing, because results are displayed and used on the fly. The software continually displays results as simulations proceed. Users benefit by receiving immediate and continuous visual feedback in a form that is understandable. In addition, the software paradigm enables complex computations to be calculated and the results intelligently processed, organized, analyzed, extracted, and displayed and overlaid in animated graphical form. The software seamlessly and dynamically merges heterogeneous and distributed geo-spatial information into a single map - joining and viewing together separate data sets that share all or part of the same space or literally “fusing” maps together. The result of this combination is a new data set that identifies complex spatial relationships that evolve over time.

The combined data may include map layers such as plume, velocity vectors, head and drawdown contours, aquifer properties and the rest of the model inputs such as conductivity, storage coefficient, porosity, and physical, geological and hydrological features such as springs, lakes, rivers, and fault lines. Data displayed may also include additional map layers such as roads, streets, and political boundaries, satellite images, and database records with street addresses or other fields that indicate a physical location, land use maps, soil types, production well locations, basin delineation, water quality, recharge, evapotranspiration, etc. available in the form of GIS coverages. This is achieved by incorporating in the software system a built-in GIS control. The use of GIS allows a user to take advantage of the vast quantity of data available today for environmental and water resource applications.

We have emphasized real-time visualization and data and results overlays because how information is presented has an enormous effect on how a modeler can understand it and exploit it for his or her needs. Humans have five senses, but only our sense of sight has sufficient bandwidth to permit conveyance and interpretation of the immense amount of data obtained from large-scale simulation applications. The new software system is highly effective since it recognizes this physiological constraint, human perceptual and cognitive factors, and takes advantage of human’s special capability to recognize patterns and images.

Real-time and On-Line Interactive Modeling.

As changes in parameters become more instantaneous, the cause-effect relationships within the simulations become more evident. A user would naturally want to interact with modeling process in real-time, right on the spot.

As stressed in the Visualization in Scientific Computing (ViSC) workshop in 1987,  scientists and engineers not only want to analyze data that results from computations; they also want to interpret what is happening to the data during computations. Researchers want to steer calculations in close-to-real-time; they want to be able to change parameters, resolution or representation, and see the effects. They want to drive the scientific discovery process; they want to interact with their data. While this would be the preferred modus operandi for most computational scientists, it is not the current standard of scientific computing.

The new software environment provides this unique real-time interactive and capability by writing the entire program in a visual interactive language. The software is written in Visual Basic that calls a number of external DLLs, including several research-oriented Fortran modules and a graphics library called Olectra Chart. The Fortran DLLs are used to perform intensive finite-difference modeling computations. The Olectra Chart provides real-time animated visualization and analysis of system dynamics.

Unlike most groundwater modeling technologies, the new software limits the Fortran DLLs to the innermost loops of the overall program (or one time step of flow or transport simulation). The visual Basic main program controls the overall programming logic, such as time stepping and nonlinear iterations, and coordination among different components, in addition to providing graphical inputs and outputs. The result is a complete point and click software environment that provides efficiency and maximum interactivity, transparency, accessibility, user control and capability of real-time visual simulation and analysis. Groundwater modeling within the new environment becomes essentially a process of high-level interactive graphical modeling and conceptualization, as if one is drawing a picture of the site. It becomes a process of pointing and clicking to delineate the area of interest, the spatial coverage of the rivers, lakes, and wetlands, land uses, aquifer materials and properties, contamination sources, and to interactively input boreholes and wells to define aquifer framework and stresses. The user can pause at any time, including during the simulation and analysis, to interact with any aspects of the modeling process. Specifically, the new real-time interactive environment allows a user, at any time during the modeling process:

q       to graphically modify domain configuration, interactively change conceptual assumptions, aquifer structures, properties, stresses, and parameters independent of spatial and temporal discretization;

q       to adaptively change numerical parameters such as time step, grid spacing, solution methods, solver parameters, and spatial interpolation techniques, without having to restart the integrated simulation;

q       to introduce contamination from different sources and initiate particle tracking and contaminant transport modeling. Transport modeling is automatically activated when a contamination source is introduced. Flow is modeled only once and automatically skipped in the second time step if it is at a steady state;

q       to “zoom” in one or more sub-areas and investigate local detailed dynamics by initiating one or more coupled nested subscale flow and transport models that run in parallel with the parent model. Boundary conditions for the local models are extracted from the regional/parent model dynamically at every time step online and automatically;

q       to activate stochastic Monte Carlo simulation, examining the impact of heterogeneity, data limitation, and uncertainty. The probabilities at any interactively specified monitoring well and spatial statistics (means and standard deviations) are computed on-line recursively and visualized as the simulation proceeds. At any given time, best available probabilistic characterizations are presented and is recursively improved as the number of realizations increases;

q       to perform, overlay, and visualize drawdown distribution, well influence areas, the area of contribution, and wellhead protection areas and to compute and instantly visualize solute mass and water budget balance over any interactively specified zones, seepage and solute flux distribution along a user specified polyline, head hydrographs and concentration breakthrough at a monitoring well; and,

q       to customize the visual presentations and manipulate the results in real-time (e.g., changing the display mode, contour, vector, and line types, legends, levels of detail, ways of presentation, the number and order of parameter layers to be visualized), and thus help us to understand the results (especially the complex interrelationships) ‘on the fly’.

Real-time Steering.

Real-time interactive modeling and visualization makes the scientist an equal partner with the computer to manipulate and maneuver the visual presentations of the abstract results. It allows scientists to communicate with data by manipulating its visual representation during processing. The more sophisticated process of navigation allows scientists to steer, or dynamically modify computations while they are occurring and directly control the execution sequence of the program. These processes dramatically reduce the time needed for conducting a model-based simulation and analysis.

At the beginning of the simulation before there is any result generated, a few important feedbacks often significantly help in choosing correct parameters and initial values. One can visualize some intermediate results and key factors to steer the simulation in the right direction. With computational steering, scientists are able to modify parameters in their systems on-line as the computations progress, and avoid something being wrong or uninteresting after long hours or days of expensive computations. Major conceptual errors can be identified and fixed very early before many dependencies on the flawed concepts are created, resulting in higher-quality model representation in less calendar time. Real-time steering can be considered as the ultimate goal of interactive computing.