This thesis has detailed the development of the Internet Watershed Educational Tool (InterWET), using the Spring Creek Watershed of central Pennsylvania as a case study. The InterWET web site is currently available through the Penn State Agricultural and Biological Engineering Department’s home page at http://server.age.psu.edu/, under Research in the area of Natural Resource Conservation and Management. Additionally the materials contained on the web site are available on a computer disk contained in Appendix D. Each chapter in this thesis has covered unique aspects of InterWET’s development and has provided different insights into creating watershed educational and modeling tools for the Internet.
Chapter 2 presented the educational theory and research behind InterWET. The subject matter for InterWET is focused on water resource issues at the local government level. To be successful, educational tools must be relevant and useful to their intended audience. InterWET was designed to encompass the specific issues of the Spring Creek Watershed, but the general categories to which these issues belong are relevant for many smaller watersheds. All local governments deal with the issues of water quantity and quality, land development and protection, and cooperation and education. InterWET is intended to help people become better informed about these issues and to better understand their role in affecting water resources. Information on water resources not only includes scientific knowledge, but also discovering how different perspectives affect understanding.
Chapter 2 also detailed how this educational theory was implemented in the design structure of InterWET. To address the issues, InterWET used the concept of microworlds to present interactive learning experiences. These microworlds (brief descriptions in Table 2.2) were arranged in sequences that helped emphasize the similarities and differences between water resource issues and perspectives. Appendices A-C give full details for each microworld and Appendix D contains the files used for the InterWET web site.
6.2. Techniques for Internet-Based Hydrologic Modeling
Chapter 3 detailed techniques such as Hypertext Markup Language (HTML), Common Gateway Interface (CGI) Scripting, JavaScripting, Java, and Third Party Software that can be used to expand hydrologic modeling to the Internet. The theories and designs of Chapter 2 rely on these techniques to make InterWET possible. Each technique has unique capabilities (with advantages and disadvantages versus other techniques), and these were discussed. The major limitation of all techniques is the difficulty of file input and output, which provides challenges for directly applying existing hydrologic models to the Internet. Innovative ways for data input and output therefore must be used. InterWET uses calculators and map interfaces to enter input and display output. Other modeling sites, shown in the examples in Chapter 3, make use of CGI scripts to deliver customized information to a client.
Chapter 3 showed that Internet-based hydrologic modeling is in its infancy. Techniques that are server-side, such as CGI Scripting and Server-Side software, are restricted by server memory, speed, and security issues. Client-side techniques, like JavaScript and Java, are relatively non-standardized, making them difficult to use for model development and testing. The examples of hydrologic modeling given in Chapter 3 show how specific applications overcome these problems by using combinations of server-side and client-side techniques and making use of pre-existing input file sets.
While successful, the examples of current Internet-based hydrologic modeling bring out the limitations of current modeling techniques. However, these techniques are continually changing and improving to address client and server needs. In the future, most techniques described in Chapter 3 will incorporate better file input and output methods. JavaScript and Java may both incorporate features like file storage and retrieval with the client's computer. Better, easier to use design software may be developed for ActiveX and Plug-In components, allowing them to have more flexibility for hydrologic modeling. Most modeling applications will continue to use a combination of client and server-side techniques, probably with client-side techniques handling file input and output and server-side techniques handling computation and database retrieval. All these improvements will help Internet-based hydrologic modeling mature from it’s current infant state.
6.3. Using Artificial Neural Networks for Hydrologic Modeling
Use of Artificial Neural Networks (ANNs) may help to make modeling on the Internet faster. Chapter 4 showed how InterWET implemented ANNs to predict changes in water resource loads caused by a selection of specific policy options. The Generalized Watershed Loading Functions (GWLF) Model was used to predict the changes in the average annual levels of nine different water resource loads, based on sets of local government policies. One load, total runoff, was found to not be substantially impacted by policy changes used in InterWET. Changes in surface runoff, groundwater runoff, and detached sediment loads were found to be affected only by one of the four policy areas, so they were modeled in InterWET with look-up tables. Delivered sediment, dissolved nitrogen and phosphorous, and sediment nitrogen and phosphorous were affected by most of the policies, and were modeled in InterWET with Forward-Backward Propagation ANNs. After initial testing to determine the appropriate network parameters, architecture, and training routines, ANNs were trained with GWLF predictions for each of these five loads. Using a set of testing data, this training was performed until each network had less than 25% error for predictions. These trained ANNs were then implemented as JavaScript calculators as a part of InterWET. Use of these calculators allowed prediction of water resource load changes at least 30 times faster than running GWLF.
The combination of ANNs and JavaScript, designed for the Internet, was successful at producing calculators that closely replicated GWLF predictions of water resource load changes, while also being much faster. InterWET provides one example of combining Internet modeling techniques and mathematical modeling tools, like ANNs, for hydrologic studies and modeling. Other tools, such as fuzzy logic or Java, may extend the capabilities for hydrologic modeling on the Internet. As Internet speeds increase and techniques and tools become more advanced, hydrologic modeling on the Internet will become easier.
Chapter 5 presents a user’s guide for InterWET. InterWET consists of 25 web pages which support the microworlds constructed for the different combinations of water resource components and perspectives. To go through the InterWET web site, one first chooses whether to follow the water resource sequence or the perspectives sequences. Both sequences go through all of the web pages, but use a different order to place emphasis in individual water resource components or perspectives. Next, the learner reads through several web pages which introduce the water resources and perspectives. Then, the learner goes through each of the web pages that support a microworld. The microworld interfaces are either interactive calculators or maps. Each of these web pages focus on one combination of the 5 water resource components and the 3 perspectives and contain brief lessons which instructor the learner how to use the microworld and supervise experimentation with the microworld to improve the learner's understanding. After going through all of the web pages for the microworlds, InterWET has a final web page which summarizes all lessons and provides links to other water resource web sites.
The two most important ideas to remember when using InterWET are: 1) InterWET was developed as a research project and 2) InterWET was designed as an educational tool. As a research project, InterWET was meant to show how hydrologic modeling could be used for watershed education at the local government level. This thesis has shown the educational designs and theories, Internet modeling techniques, and computational methods that can be used to do this. However, as an actual software product, InterWET is at a "Beta" testing stage, meaning that the InterWET web site has not been evaluated beyond the focus of this research, and it is known that there are currently some browser capability problems. Due to the time limitations inherent in research projects, no studies were performed to evaluate the educational effectiveness of InterWET’s layout, microworlds, and web page lessons. In addition, the 32-bit versions of Netscape 4.0 and 4.5 for Windows are the only known web browsers on which all web pages and microworlds in InterWET will correctly run. Some brief testing has shown JavaScript compatibility problems with Internet Explorer browsers, JavaScript memory problems with 16-bit Windows Netscape Browsers, and Java compatibility problems with Macintosh versions of all browsers. As a research project, the research focus of InterWET was to show what was possible, not to make bug-free, commercial quality software.
The second idea to remember when using InterWET is that the web site was designed for educational use. InterWET uses established hydrologic modeling and computational methods, but also many simplifying assumptions. These assumptions were made to bring out the most important underlying concepts for each microworld. The values predicted by each microworld are not nearly as important as the concepts implied by these values. For example, the level of numeric increase in surface runoff caused by selecting wetter soils is less important than the concept that wetter soils produce more surface runoff. The values predicted by the microworlds and listed in tables on InterWET web pages should not be used for actual design and decision making. Much more detailed analysis should always be performed when new structures or policies are being considered. This thesis and InterWET do contain numerous references to original data, such as stream flow and groundwater flow levels, that should play a role in decision making. InterWET was not designed to replace the work of scientists and engineers in watershed planning, but rather to help others understand their work better.
Even though InterWET may not have the polish of a professional software product, it does provide many valuable contributions to society. Specifically, InterWET can contribute to the Spring Creek Watershed Community, the Educational Community, and the Scientific Community. For the Spring Creek Watershed Community, InterWET can act as an educational resource for local government officials and concerned citizens. There are plans to give a formal presentation of InterWET to some of these decision-makers, who can use InterWET to better understand how their decisions affect water resources. They can see the difference between each perspective and water resource component and better grasp the complex interrelationships between hydrology and policies. Beyond decision-maker education, government and conservation groups can use InterWET in existing programs to educate other local citizens. As information is being gathered on Spring Creek Watershed, InterWET can also act as a foundation for education and information delivery.
While InterWET is valuable at a local level, it also contributes to a much larger Educational Community. As a stand-alone, self-guided learning environment, InterWET can be used by anyone interested in learning more about water resources. High school and college courses can use InterWET as a supplement to environmental studies; to help amplify the important role that computer modeling plays in environmental research and planning. Furthermore, InterWET can act as a blueprint for computer- and Internet-based educational modeling. Not only does InterWET demonstrate a structure based on educational theory, but also shows how different Internet modeling techniques can be used to produce microworlds. This contribution will come primarily through publication of Chapter 2 as an article in an educational research journal.
Moreover, InterWET can contribute to the Scientific Community. InterWET is one of only a handful of hydrologic models available via the Internet. As noted in reference to the Educational Community, InterWET provides examples of how new modeling techniques like JavaScript and Java can be employed for computer modeling on the Internet. Also, InterWET can introduce scientists to the microworlds concept, which provides an established basis for converting hydrologic modeling (for the traditional scientific role) to an educational role. It is intended that Chapter 3, describing Internet-based hydrologic modeling, will be published in a general, hydrologic scientific journal so as to reach a wide audience.
Last, InterWET shows scientists how artificial neural networks can aid Internet-based hydrologic models to greatly increase the speed of calculation. As with Chapters 2 and 3, plans are to publish Chapter 4 in a research journal; one that will reach an audience interested in both hydrologic modeling and artificial neural networks. The combination of hydrologic modeling with Internet modeling and with artificial neural networks will hopefully show the Scientific Community a new approach that can have numerous applications in many fields of research. Through public presentations and publishing of Chapters 2, 3, and 4, InterWET can make valuable contributions to assist Spring Creek citizens, educators, and scientists.
6.6. Future Work Based on InterWET
This thesis discusses the formation of InterWET and demonstrates its use. Just like the field of Internet-based hydrologic modeling, InterWET is in its infancy and will require work beyond the scope of this thesis to fully capitalize on its potential. The current version of InterWET should be evaluated by several groups. Both educational and Spring Creek experts could provide valuable feedback on the design of the web pages and microworlds. To fully evaluate the educational potential of InterWET, research should be performed to assess if InterWET increases the knowledge of a variety of learners. This research could also highlight what elements in InterWET could be modified to increase understanding.
Beyond evaluation of the present model, future research could modify the current version of InterWET. InterWET could be updated with newer Internet modeling techniques or expanded for different water resource components. Modified web pages and microworlds could be developed so InterWET would work completely on most, if not all, web browsers. InterWET could also be adapted to other watersheds beyond Spring Creek. Even with these enhancements, InterWET should remain primarily as an educational tool.
From the local government level, InterWET could become an integral part of an education and information delivery web site. This web site could be based around InterWET and include new features like real-time stream flow maps, more detailed models, and reports and data on specific sites. Also, the different microworlds in InterWET could be improved by utilizing results from more detailed studies on Spring Creek Watershed.
Modification of the current InterWET will only be a small part of the future work based on this research. Most of the future work will come from research by educators, scientists, and engineers inspired by the example of InterWET. The contributions InterWET makes in the educational and scientific communities may lead to work and research in areas including educational computer models, Internet-based hydrologic models, educational hydrologic models, and Internet-based artificial neural networks. As is often the case, this research probably raises more questions than it answers, but that keeps the work in this area ever expanding.
Future researchers in this field should keep several things in mind. First, as use of Internet delivery increases, other researchers in disciplines outside of hydrology will also be tackling the same kinds of problems as faced in this study; such as file input/output, user interfaces, memory, and security. Instead of waiting on published literature, hydrologic modelers will need to search the Internet to stay up-to-date on the advances in Internet-based modeling. The methods that work for current technologies and modeling techniques may become quickly outdated. Web-surfing will be a valuable part of determining the current state-of-the-art.
Second, future researchers will need to rethink the traditional pattern of the separate steps used in hydrologic modeling; making an input file, running the model, and evaluating the output. To increase speed and to appeal to a wider audience, Internet-based models will need to combine these steps utilizing features like more interactive graphic user interfaces, simpler input and output procedures, and modification of the interface according to user preferences.
Third, the idea of making a model, sending off copies, and moving on to the next project will change. Any successful Internet-based model will need continual maintenance, updating, and user support to adjust to changing technologies and web browser versions. While the Internet provides an ideal medium for model delivery and updating, it also requires modelers to address issues of security, memory storage, and delivery speed. Modelers will need to anticipate, as much as possible, evolving Internet structures that affect model performance.
Last, the future Internet-based hydrologic modeler should not be afraid to experiment. Many techniques for Internet modeling are so new that few people or books fully know all their features. In some cases, like JavaScript's memory restrictions on 16-bit Netscape Windows browsers, these features may restrict possible applications. In other cases, like newer versions of Java using client-based files, these features may greatly expand the capabilities for hydrologic modeling.
The Internet Watershed Educational Tool (InterWET) provides a unique
approach to hydrologic modeling using the Internet while integrating educational
objectives. Based on sound educational theory, this tool uses the latest
mathematical and Internet modeling techniques to deliver sound hydrology
that is easily understood by conservation managers, decision makers, and
the general public. It provides a vision of future applications of hydrologic
models.