Model Integration

hydrologic modeling:

model integration

Model Integration
Critical Success Factors
Key Task Elements
Design of Integrated Modeling Environment

In water management arena, models are abundant in numbers as well as in categories, ranging from hydrology to water quality, economics to ecology. Due to this multi-disciplinary nature of water management, a need for automated or semi-automated connectivity among models of different functions and types is necessary for rapid alternatives evaluation. In addition, an integration of models increases the value and reliability of information by

providing easy access to data and results; and
ensuring data integrity through a common data platform.

Models are inherently connected with data and maps. Hence an integration of models also implies integration of data and maps. As a result, an Integrated Information Environment (IIE) can be considered as an umbrella environment where models, maps, data, and expert judgment, which are used in making decisions can be integrated. It should be noted such integration environment are also called IME (Integrated Modeling Environment); IDE (Integrated Decision Environment); or DSS (Decision Support Systems).

Integrated Information Environment (IIE)

Click on this image to see an example of Model Integration

Critical Success Factors:
The three critical success factors of a model integration task are:

speed of water management alternatives evaluation;
accuracy and reliability of analyses;
verifiability and quick replicability of the analyses.

Key Task Elements:
The key elements of a model integration task are:

Development and documentation of data sharing and exchanges among different models used in the project, including flow charts;
Development of a semi-automated (full automation may not be possible due to the need for modeler interpretation and quality control checks) processes and linkage programs for seamless linkage of different models;
Development of necessary visualization tools and quality control tools and processes to ensure that the data exchange among models are accurate and reliable;
Rapid evaluations of alternatives and variations thereof using the semi-automated tools and proper documentation of results;
Provide specific model expertise and coordinate among a diverse set of modelers with different needs to ensure (a) consistency among modeling assumptions and shared data sets; (b) accuracy and reliability of model results; (c) interpretation and presentation of results to decision makers and stakeholders;
Consistency with the data standards, adequacy of documentation, and ease of verification/replication of results;
Provide expertise for necessary model development/modifications to be able to represent an alternative with model input data;
Attention to detail.

Based on experience, it has been found that a suite of flexible stand-alone programs with minimum manual intervention provide the best opportunity for accuracy, reliability, and speed in multiple model linkage situations.

Design of an IME (Integrated Modeling Environment):
Once the concept for the IME has been agreed upon, the next step is to develop a detailed design of the IME. This is an interactive and iterative process. The design is a flow chart or layout of the entire IME. It shows all the selected models and their databases, their interaction with each other, the graphical user interface (GUI) for running the models and accessing the databases, the graphics tools for visually displaying data and model results, and editing tools for viewing and editing data and model results. It is desirable to start with a preliminary design and gradually develop the final design to ensure that all elements necessary to make the IME functional are included.

The design also must show any FORTRAN/C-shell programming necessary to implement the IME. Because, without powerful programming links or hook-ups among models and data, the attempt to develop an IME would simply fail. It is our opinion that without an outstanding programmer in the design and programming team, implementation of the IME is impossible, because the power of the IME lies in the infrastructure of models, data, and their links, not in how pretty the GUI is. On the basis of the design of the IME, at this stage, all the identified FORTRAN/C-shell programs are written and thoroughly tested to ensure that they perform flawlessly. This is not an easy task and it requires vision, creativity, and persistence.