About Electronic Atlases
Original author, P. Rhines; revised by J. Swift
Today's desktop computers offer us the facility to hold and examine large
quantities of ocean data in a way that uniquely supplements paper atlases.
George Batchelor speculated (J. Fluid Mech. 1981) that computers would not
replace paper journals: "Personally I should regret the introduction of any
system which did not allow me to hold in my hand and take in a quiet corner
of my own choice, a thing like a book with paper pages which are printed in
attractively designed type." Many of us feel attached to paper for the
simple reasons of stability and annotation: it is difficult to read
something profitably without marking it all up, and one wants it to be
around for at least ten years, probably more. But much of the hard-won
accuracy and breadth of ocean data is lost on the way to the journal page.
The way we form ideas and paragraphs with a fast word processor is
remarkably different from the original pen and ink process. The computer,
once just a calculator, has enabled us to work in new ways. Similarly, the
computer can enable us to examine data in new, flexible ways, and, rather
than replacing traditional atlases, permits us to take new directions.
While traditional temperature and salinity measurements remain the heart of
physical oceanography, we are rapidly being drawn into areas where new
variables are important: trace chemistry, biology, three-dimensional
distributions of atmospheric variables and sea surface imagery. This surge
of data requires a new approach. Perhaps the oldest problem in oceanography
is the unanalyzed or under analyzed data set. This grew far worse with the
massive effort which went into ocean data collection in the 1990's. A
modest contribution toward solving this problem would be a simple means by
which a new ocean section could be sent through the mail and immediately
viewed, 'picked up, and turned over' using widely available hardware. That
is one intent here. Another intent in this work is to provide an
educational tool for ad hoc real-time visualization of oceanographic data,
to permit casual, easy exploration of the oceans at low cost and in new
ways.
Stommel and Fieux (1978, Woods Hole Press) presented a Guide Michelin of
ocean atlases. This little book summarized the paper atlases available
then. Most oceanographers spend time with these atlases and they are
wonderful to have at sea on a cruise, to provide a context for a new
section or perhaps a new type of measurement that the authors of the atlas
never could have envisaged. But they are frustratingly rigid in their
presentation, preventing us from making new connections and comparisons by
simple scale changes, or changes of plot variable. Many oceanic variables
have been measured on cruises like the Transient Tracers in the Oceans
(TTO), South Atlantic Ventilation Experiment (SAVE), and, especially the
World Ocean Circulation Experiment (WOCE). The potential number of
property-property plots with, say 15 variables, is large, and with plots
using 2 dimensions plus color, or even 3 dimensions plus color, we have
nearly limitless possibilities.
Computers have been used to plot and analyze data since their inception (an
example is Fofonoff's early (J. Fish. Res. Bd. Can, 1962) analysis of
northeast Pacific wind-stress curl which, he admitted, was initiated to
find a use for the new computer at Nanaimo: he may be the first
oceanographic computer junkie). Individual groups with a collection of
stations have often produced efficient plot software. Those who have major
ocean-going projects have thus developed their own systems. Some of these,
like the acquisition and plot programs of the Oceanographic Data Facility
group at Scripps, or the section contouring applications developed by Dean
Roemmich and Lynne Talley, are remarkably facile. These systems are
currently on Sun desktop systems. Electronic atlases are a slightly
different concept. Tom Sgouros and Tom Keffer at Woods Hole in the mid
1970's wrote a program called CAMS-ATLAS for VAX machines. This program
allowed the user to assemble data from the attached disk archive, and plot
and contour sections, property-property plots, map variables on a horizon
of choice (like sigma-theta). Part of the goal was to make the first
extensive set of potential vorticity maps.
These and other 'expert' systems, typically on UNIX workstations, remain a
key tool of the deep-sea oceanographer. Nevertheless there are many who do
not spend their full time overseeing such data sets, yet need to use them
just as one needs an atlas. They (and perhaps even the experts) need a
facile way of participating in the stream of new data, and in trading it
around. And one wants the data 'in entirety', without degradation of
accuracy, not an internal report synopsis or the output of an inverse
calculation of uncertain quality. We at last have been given the power to
examine hydrographic and tracer data without building a massive effort of
computing and plotting.
These matters must have been in the minds of Stommel and Luyten (WHOI Tech.
Report, 1989), when they produced a set of North Atlantic data on floppy
disks. They combined the data set with an elegant 120-line Basic program,
PLOTTER.BAS, to view them.
In 1989 the program ATLAST [for IBM-PC's] took the example of these earlier
efforts, and developed the goal of 'browsing' the data, zooming in so that
the full accuracy can be accessed, and replotting in ways which are
natural. Its object was not only to improve the convenience of conventional
data plotting, but also to suggest new kinds of analytical plots.
ATLAST stimulated the development of OceanAtlas for Macintosh, later called
Power OceanAtlas, which works with ocean section data. OceanAtlas provided
a set of simple tools to examine on screen whatever data facets and
features are of interest to the examiner. OceanAtlas linked the browsing
features of all plots. Like ATLAST, OceanAtlas was not a data presentation
program, but was more nearly an ad hoc real-time data exploration program.
Reiner Schlitzer's program Ocean Data View [for IBM-PC's running Windows]
includes a wide array of property-property, section, and areal plots, and
supports hardcopy output. It is somewhat less oriented toward data
exploration than are ATLAST and OceanAtlas, but can produce stunning plots.
Thanks to generous support from the US National Science Foundation and the
US National Oceanic and Atmospheric Administration, most features of
OceanAtlas have been moved into a Java application, Java OceanAtlas, and
many new features were provided. Java OceanAtlas has the advantage of true
cross-platform compatibility. Although a separate installer in required
for each operating system, the core files are literally the same for MacOS,
Windows, and UNIX systems, making application and feature upgrades a simple
matter.
By providing a true multi-platform data exploration environment with a
large, vetted, diverse data set, we have now come quite close to our vision
of a true electronic oceanographic atlas.
We hope that you enjoy using Java OceanAtlas and find our data selections
interesting and educational.