CALMet99 Presentation
The Development of VISITview
by Tom Whittaker and Scott Bachmeier
Cooperative Institute for Meteorological Satellite Studies
University of Wisconsin-Madison
Madison, Wisconsin, U.S.A.
Abstract
The evolution of a distance-learning, collaboration tool named VISITview,
which is being developed for use by the VISIT (Virtual Institute for Satellite
Integration Training), will be discussed. Both technical and functional
aspects will be described, along with observations about the efficacy of
this development approach. Trade-offs between buying commercial software
and developing on one's own will also be discussed. Portions of one remote,
Web-based lesson will be used to illustrate these points.
Introduction
The Virtual Institute for Satellite Integration Training (VISIT) is a collaboration
between NOAA Cooperative Institutes (CIMSS,
CIRA
and COMET)
to produce training materials and provide lessons for U.S. National Weather
service personnel. This effort is part of a much larger national
program, but focuses on the use of integrated multi-sensor data in the
analysis and forecast process. During the past year, we have been developing
a software tool to enable this training to be done remotely ("teletraining"
is the term used to describe this). This has become known as VISITview.
VISITview is a Java based, distance-learning tool that uses a client-server
broadcast model. Instructors produced lessons using GIF or JPEG images
and provide specific information about these via HTML. Lessons are
taught in an interactive manner using the telephone for audio and VISITview
for keeping everyone looking at, and interacting with, the same scenes.
The motivation
In early 1998, I was approached by Scott Bachmeier, an instructor for VISIT,
about the availability of tools that might have better applicability to
his particular training needs that the system currently in place.
The latter is a commercial product that requires a combination of software,
a PC running Windows, a specialized modem to multiplex voice and data,
and a speakerphone. The main aspects of this system he was finding
awkward was the inability to have a continuous loop of images and to annotate
over more than on image at once. While he also did not like the duplex
voice/data, his main emphasis was on the functionality.
I investigated commonly available Web-based collaboration and distance
learning tools. Since the Weather Service was also in the process
of deploying the AWIPS system at the time, we felt it essential that a
platform neutral solution would be preferred. What I found was a variety
of "whiteboarding" and "chat room" applications, in addition to some sophisticated,
business oriented conferencing software. None of these really addressed
the problems he described.
The start
At the time, I was actively involved in the development of Web-based tools
for showing animations of
satellite images and decided that it would not be difficult to modify
this software to accept its commands to state/stop/step the animations
via the Web. Using a public domain framework for a broadcast client/server
model, I spent a single day producing a sample for him. After a few
refinements, we decided to involve others in this to discuss whether this
approach would merit continued development.
The crucial test at this time showed this model could support tens of
users simultaneously. It is worthwhile to note that at this point,
the client software was written in Java as an applet so it could run in
a browser without requiring any new software to be installed on the students'
machines.
More functions
The early animator soon had the ability to draw lines. It was important
that these annotations show up on every frame, so consideration was made
to provide a pleasing display that did not distract from the instructor
or students' learning. In a flurry of enhancements, we added:
-
wide lines for annotations
-
rocking of the animations (in addition to the movie loop)
-
changeable background color
-
a 'portal' into another image with common cursor positioning
At this time, we also began a series of live tests with Weather Service
Offices, and got out first big setback: the variability of connections
meant that some offices were getting images and commands in a timely manner
while others were not.
Teletraining stresses the communications in two ways: first, many
image files need to be transferred from the server to the clients in realtime
which means a high bandwidth is needed. Second, command information
(start/stop loop, step frame, etc.) require fast response time. What
we found very early in testing was that the connectivity of offices to
the Internet was so variable that it was impractical to do both during
a training session.
We also had enough experience by now to know that the functionality
of this tool was just what was needed for the VISIT program training.
The obvious approach at this point was to have offices download the image
files prior to the training and then use the applet to read from the load
disk. Java security, however, prohibits an applet from doing this,
so we were forced to either go the route of purchasing a signing certificate
or distributing an application which was run locally but controlled remotely.
Since the largest volume of data in a lesson is the image files, we
decided to just put a wrapper around the applet that would allow it to
run as an application (and thus be permitted to read the local harddrive
and communicate over the Internet to another site).
This approach substantially improved the situation; however, as I'll
mention later -- it is still not a perfect world.
As the instructors gained more experience and received feedback from their
lessons, a barrage of changes were made to what had now become known as
VISITview. These included:
-
local or remote files
-
show status information for all students
-
add groups to the server so only one server is needed at a site
-
permit fading from one image to another
-
add color enhancements during the training
-
allow for multiple "portals" and a transparent color level
-
connect to the RAMSDIS On-Line database to test collaborations with realtime
data
The last bullet is quite an interesting application. The RAMSDIS
On-Line web site provides access to realtime satellite products through
CIRA. With their help, we created a
VISITview "lesson" that consists of the scenes from their database.
Thus, it is possible for two remotely located people to share a view of
data and, using the telephone, discuss what they see. The model for
this is that anyone connecting is a "master" (instructor) and thus can
control all aspects of the session.
Making VISITview available
During this time, I was encouraged by the Unidata User Committee members
to make VISITview available to the university community. CIMSS decided
that this was an excellent idea, and in late 1998, a VISITview
Home Page was created and documentation supplied to allow anyone to
get the tools and use them. Our vision was that labs with many computer
workstations would benefit from this.
During the month after the announcement was made to the Unidata universities,
more than 100 individuals took a look, and about 20% downloaded the software.
What's next?
As of this writing, the list of enhancements from the instructors includes:
-
being able to link any page of the lesson to a URL
-
a lesson builder tool to make putting together a lesson easier
-
text-based discussions via a 'chat window'
-
quiz questions with student responses displayed on the instructor's console
Risk factors
It is important during the development of any software to be used by an
organization to assess the risks involved. The Weather Service Training
Branch is constantly monitoring and evaluating for all their efforts.
I want to talk about two sides of this as they relate to VISITview:
-
Control Who controls the evolution of function
of software? Generally, the one paying for it. Although this
is not strictly true, in some cases the needs or desires are so specific
that a commercial vendor may not want to incorporate highly specialized
function that they will then have to support.
-
Viability In the commercial world, a product is
occasionally discontinued. If you purchased this and want to continue
to use it and have support, you need to then seek another option.
Patents and copyright laws may legally prevent you from doing this, and
it may be so costly to try to support this within your organization that
you might have been better off doing something else.
We believe we have minimized the risk factors and maximized utility by
doing this development in a closely coupled, spiral type process.
By involving more than one software developer, we have also started to
insure that this can be maintained and enhanced beyond the involvement
of the original people.
Final comments about teletraining
There are several non-technical issues related to teletraining. While
I do not want to dwell on them, I would like to highlight a few of the
closely related concerns. First is efficacy. Our goal is to
increase a forecaster's ability to do his/her job. "Ability" may
mean efficiency, but it may also mean accuracy. Either of these is
difficult to assess -- especially over the short term. Common measurements
of the effectiveness of learning, namely knowledge or skill testing, does
not usually measure the long-term impact. In addition, teletraining is
just one of several different approaches to training being used, and mainly
as a substitute for some of the in-residence training (because of cost).
The second is related to pedagogy. The very nature of Distance
Learning means that some of the techniques used during in-residence training
need to be changed. For example, more emphasis must be placed on
vocal interactions since other cues are not available to either the instructors
or students. In addition, more overt effort is needed to keep the
students engaged; in some cases, this means different techniques are essential.
There are other issues related to teletraining, which I see mainly as
challenges. Those of us involved on both the technical and pedagogical
side look forward to meeting these.