Abstract

Teachers in many parts of rural Colorado have an interest in adopting GIS technology but lack technical resources and expertise to set up and maintain the application software. This paper reports a campus showcase project to implement Internet-based GIS training modules for distance learning. The modules are embedded with online assessment tools, allowing a dual project focus on GIS content and on strategies for automated user assessment. The modules are delivered on an Internet Map Server, and will be assessed in lower and upper division college classes. The paper will report on the mechanics, the module content and the assessment strategies.
 
 

Introduction

Teachers in many parts of rural Colorado have an interest in adopting GIS technology but lack technical resources and expertise to set up and maintain the application software. The University of Colorado at Boulder has initiated a campus-wide initiative. The Alliance for Technology, Learning and Society (ATLAS, http://www.colorado.edu/Committees/ATLAS/) has the intention of utilizing information technology generally, and the Internet specifically, to deliver educational resources and services to the community. The term 'community' in this context is writ large. For example, communities may be local neighborhoods or K-12 schools, the Boulder or Denver business community, or the community of Colorado citizens who might attend or send their children to the University. One primary service includes distance learning.

This paper reports an ATLAS Showcase Project to implement Internet-based GIS training modules for distance learning. This is a one-year campus project, demonstrating how GIS can be delivered to any computer running Netscape or another Internet browser. Additional impetus for the Showcase Project driven by a growing on-campus demand for access to GIS software for classroom use by faculty in Anthropology, Environmental Biology, Geology, and other earth science departments. These faculty want to incorporate GIS into their classrooms, but lack have departmental resources to administer a large GIS software package. The campus Computing Center lacked staff expertise to install distributed GIS services. The Showcase Project implements a prototype, concurrently developing distance learning content and demonstrating the mechanics of how to deliver this content via the Internet.

Three GIS course modules are designed to teach Physical Geography undergraduates where their drinking water comes from, using ArcView GIS. The faculty teaching these modules do not have to install or manage a large GIS software package, and can focus their teaching on a topical domain (in this case, mountain hydrology). Faculty will be able to create their own modules and submit them to the IMS site, to tailor exercises to their own class material. The modules are designed and ready to be tested in undergraduate Hydrology and GIS classes this fall. Following the testing cycle, arrangements will be made to incorporate these and other GIS modules into rural Colorado classrooms.
 

ATLAS IMS Website and Architecture

The architecture includes three components, including an Internet Server, which is Netscape FastTrack, running on a Sun Ultra 2 with dual 200 mHz processors and Solaris 2.5 UNIX operating system, http protocols and CGI utilities. The IMS server is a Gateway 6000 platform running multiple ArcView 3.1 sessions on a 266 mHz Pentium II processor and Windows NT 4.0 oeprating system. Multiple sessions were found to speed performance when more than three users accessed the IMS server concurrently. Clients could operate Netscape 4.5 or Internet Explorer 4.0 on Macintosh, Windows, AIX or UNIX platforms. (Subsequent to the research reported in this paper, the client browsers have been updated.)

The ATLAS IMS Website (http://stormking.colorado.edu/atlas) includes numerous components.  The Internet Server component (running oon Stormking) actually serves the Web pages, responding to client requests.  Each Internet Client component is running an Internet Browser, in which a Web page is visible that looks like an actual ArcView session.  The user clicks on a button inthe ArcView session window.  This sends an http "GET" request back to the Internet Server, which passes the IMS requests over to the ArcView server (running on Paris).   The ArcView Server contains a delegating mechanism (commonly referred to as a "traffic cop") which mediates multiple requests to multiple ArcView sessions.  The request is delegated to the proper ArcView session, which performs the actual ArcView command, which changes the appearance of the view window (a zoom, or a pan, or an identify function is performed).  MapCafe then generates a new version of the ArcView window in http protocol, and the 'traffic cop" sends it back to the INternet server, which passes it back to the client.  CGI protocols are necessary for any client requests (such as comments, or requests for online help) that requre html forms.

The complexity of our set-up was mandated in part because of our aim to demonstrate to the campus Computer Center that an IMS server could be installed using their hardware and selected software. Another reason for the complexity is that the IMS architecture requires an Internet and an ArcView server. In most cases, these two servers could operate on a single machine. However, our computer center does not permit user access to the machines running their Internet servers, so we had to implement the ArcView server on a separate machine, to preserve our own control over installing ArcView themes and views. We believe that other campuses trying to install an IMS system utilizing campus resources will be in a similar dual-control situation, and therefore it seems informative to report on our experiences, in the context of lessons learned, if nothing else.
 
 

System Benchmarks

System benchmarks were performed in November 1998, just after the system came online. Two client environments were tested, accessing larger and smaller volume datasets (described below). All benchmarks show substantially slower response times when the client utilizes Internet Explorer 4.0, and in most tasks, these are unacceptable for preserving user satisfaction. We found that in many cases, users would terminate the Map Café JAVA applet before it had completely launched, in some frustration. Netscape 4.5 shows a faster performance, and this may be due in part to the fact that the Internet Server was running Netscape FastTrack. We did not have an opportunity to run benchmarks utilizing a similar Microsoft Internet Server product.
 
 

At present, three data sets populate the ATLAS GIS Web site. A small-scale Colorado data set (http://stormking.colorado.edu/atlas/co-1/colo1.html) has been provided courtesy of Esri (http://www.Esri.com), and includes seventeen layers of thematic, environmental, and demographic data. Layers include hydrography, transportation, landuse, census demography, recreation, public land ownership, and other information, stored as ArcView layers and shapefiles. A second data set (http://stormking.colorado.edu/atlas/co-2/colo2.html) includes public domain 90-meter Digital Terrain data (DEMs). These data were compiled by USGS (http://www.usgs.gov), and have been imported to ARC/INFO, converted to lattices, and ported over to ArcView. Layers from these two datasets were used in the benchmarks reported above.

The third layer forms the basis for the three GIS modules. These data include GPS readings for snowpack in 1997 and 1998 for the Green Lakes Valley and Niwot Ridge Research Study Areas http://stormking.colorado.edu/atlas/niwot/rockies.html). The study area extends about six kilometers east from the Continental Divide, with elevations ranging from 3,200 meters to

almost 4,100 meters. Data collection includes snow depth, latitude and longitude (in spherical and UTM coordinates), elevation in meters, and ancillary field observation of environmental conditions. These data were provided courtesy of the Niwot Ridge LTER (Long Term Ecological Research) Project centered on campus (http://culter.colorado.edu:1030/).
 
 

GIS Modules

The ATLAS IMS Website (e.g., http://stormking.colorado.edu/atlas/gis/module-1a.html) has a frame-based design. The ArcView IMS sessions run within a JAVA-based (MapCafe) frame, surrounded by several html frames providing the assignment scenario (above the ArcView window), ancillary data windows (to the right) and a thrid frame (below) where students can ask for help or make comments. The uppermost (html) frame will incorporate questions that the students will answer in short essays, via html scrolling form templates. In this way the assignment can be handed out, completed, handed back in, graded and returned via the Internet.
 

Three course modules have been implemented to prototype the ATLAS IMS Website. These modules target a lower division (freshman/sophomore) level and assume no prior knowledge of GIS. Students will take a virtual tour of the Niwot Ridge study area before beginning the exercises, visiting Web pages to view maps and photographs of the area.

The three modules follow one upon the other, and inform students that Boulder's drinking water supply comes from mountain snowpack, and how to monitor the snowpack depth to predict availability of drinking water. Students learn how to enter snowpack measurements into a GIS. Then they learn two GIS methods for generating digital terrain maps from elevation measurements, and how to determine which method is likely to be more reliable. Finally, they use GIS to merge the snowpack with the digital terrain and a landcover data theme to build a model of snowpack availability. The modules are described below.
 

GPS Module 3 - Making the Final Model
The snowpack volume in an area is affected by many environmental factors. In this exercise, you'll work with two factors, first the terrain data that you used in the previous module. A second factor is landcover, such as bare rock, soil and vegetation, talus, glacier, and so forth.

In this module you'll learn about how snow depth is associated with terrain and landcover, and make a composite GIS model of all three layers. You'll learn how GIS models are built and tested, and answer a few questions along the way. The exercise should take 45 minutes.

Assessment Tools

Usability testing forms an important component of any online service, and educational materials are especially important to test during and after the design phase (Buttenfield, 1999). These modules will be usability-tested in a Snow Hydrology class and in an Introductory GIS class during Fall semester 1999. Usability testing by hydrology students will evaluate course content, and testing by GIS students will evaluate the interface and delivery mechanisms. Assessment will include several strategies. First, students will be given an entry questionnaire before visiting the Website, to describe their background in hydrologic modeling, in using the Internet, and in working with GIS. During their work with the modules, brief focus sessions will be held where students can describe their experience, critique the Website as a group, share trials and tribulations, and learn from each other how to improve their own performance. At the end of the three exercises, students will be asked to write a short essay describing what they have learned about the source of drinking water in Boulder, what they have learned about learning on the Internet, and whatever comments they have to improve the Website. These essays will be handed in anonymously, to insure open and honest critique.

Second, the Website has three kinds of buttons embedded in each module. The Comments button allows them to ask questions, critique the site or the exercise. It is anticipated that the testing phase will require instructors to hold "online office hours" to answer student questions and respond to comments. A Help button will provide hints and short walk-through tutorials to assist students in using the ArcView GIS tools, and in using their Internet Browser software.

The third kind of button is actually a pair of buttons, including a Good and a Bad button. Students will be instructed to hit these buttons respectively if they either like or dislike what they are doing on the Website. The buttons send a message back to the instructor, alerting what Web page the student is working on. It is intended that the Good and Bad buttons will help to refine the Website design. Buttenfield and Kumler (1996) report on use of a similar tool in evaluating the Alexandria Digital Library.
 

Summary

This paper reports on design and implementation of an ArcView IMS site to deliver GIS tools to University of Colorado classrooms. The project was intended to run for a single year, and in planning, it seemed that 12 months would be ample time to deisgn the modules, implement the IMS and other software, and run usability tests in actual classrooms. That timeline has proven inadequate.

In the first twelve months were able only to implement and stabilize the system. Our biggest delays came from integrating system architectures on two separate platforms, and in creating a stable environment. The early implementation crashed repeatedly due to lost network sockets. The problem was eventually solved by moving the ArcView sessions to a larger machine, and by launching multiple ArcView sessions.

A second problem has plagues us throughout the project, namely that the ArcView IMS does not contain the full functionality of ArcView. One can pan, zoom, identify, and toggle theme visibility. To perform any other GIS functions (to bring up a table, for example, or interpolate contours) one must either write the functions using Avenue or MapObjects. We devised an alternative solution, namely to implement a series of buttons on the IMS menu bar. Each button relates to a step in the exercise. The student pushes the buttons in order (left to right). As each button is pushed, a new View is generated, showing results of what the actual GIS commands would have generated. In effect, the student is not learning how to "do" GIS, but rather working through a fixed sequence of operations to see how to "do" hydrologic modeling using a GIS package. Beginning students will probably not know the difference. However, the bottom line is that in current versions, the ArcView IMS is not suitable for teaching GIS skills, as it provides only a slide show capability. Future versions of ArcView IMS are expected to increase user functionality, and it is hoped this limitation will be remediated soon.

We anticipate testing Boulder classrooms this fall, as we design additional modules, and delivering a redesigned ATLAS IMS site to rural Colorado classrooms within the coming year.
 
 

Acknowledgments

This research forms a portion of the Alliance for Technology, Learning and Society (ATLAS) Initiative Showcase Project " Distributing GIS for Environmental Applications: An Internet-Based GIS for the Rest of Us". Funding from the University of Colorado at Boulder is gratefully acknowledged. Additional funding for data collection were provided by The Niwot Ridge Long Term Ecological Research (LTER) Program (NSF award DEB-9810218), with funding from the National Science Foundation.
 
 

References

Barbara P. Buttenfield and Ming-Hsiang Tsou
Department of Geography CB-260
University of Colorado
Boulder, CO 80309
babs@colorado.edu, tsou@spot.colorado.edu
http://greenwich.colorado.edu