Transcript Slide 1

Using Kriging and Interactive Graphics in Web-Based Application for Spatial-Temporal Trend
Analysis of Ozone and Weather Systems in Central America
Kening Wang, Charles Stegman, Sean W. Mulvenon, and Yanling Xia
University of Arkansas, Fayetteville, AR, 72701
Abstract
This study illustrates a demo system which was
designed for supporting web-based interactive
visualization and spatial-temporal trend analysis of
ozone and weather data in Central America.
Longitudinal prediction maps of ozone and surface
temperature were generated using Kriging method.
Bar charts of air pressure, cloud coverage, and
elevation at each measured location were plotted on
the maps. Prediction maps, time-series chart, and 3-D
graphics are linked and presented through this
system. Through building up a relationship between
maps and other graphical representations, users are
able to effectively explore the data in many different
ways and from many different angles.
Results
Users can view the longitudinal prediction maps of ozone
and surface temperature.
Users can view graphics of elevation.
To illustrate the applicability of the proposed approach,
we have developed a full-featured demo system at our
web site: http://normes.uark.edu/ASA_Comp/.
Users can view graphics of air pressure.
Introduction
Environmental data collected by federal government
agencies usually are spatial-temporally referenced
massive datasets with multivariate measures covering a
huge geographic area. Effective visualization and
exploration of this kind of datasets create greater
challenges. Web-based data visualization can provide
users with high user interactivity and it is likely to be
very useful for exploring this kind of data. Meanwhile,
use of the internet for delivery of data and information
also enables a large amount of graphics to be accessible
to the public, and encourages users to explore the data
in a playful way.
Users can view 3-D graphic of globe, which shows the
geographic area data collected, and project description
when they clicking the tab “MY PROJECT
OVERVIEW”.
Users can view graphics of cloud coverage.
Objectives
1) To provide a graphic summary of important features
of the data, and to make the results easily
understandable to publics who are not familiar with
using geostatistical methods to generate prediction
maps.
2) To demonstrate effectiveness of web-based data
visualization for exploiting the important relationships
between the variables and for revealing the spatialtemporal trends.
Method
Kriging method has been reported to be used for ozone
prediction (Lefohn et al., 1994). In this study, prediction
maps of ozone and surface temperature were generated
using Kriging method. Three frequently used models in
this study for Kriging are: 1) Spherical model, 2)
Exponential model, 3) J-Bessel model.
The software ESRI ArcGIS Geostatistical Analyst was
used to create prediction maps, because Geostatistical
Analyst bridges the gap between Geostatistics and GIS,
and it contains a series of easy-to-use tools.
The software Nvu, which is a complete web authoring
system, was used to develop the web of this project.
Users can view Time-series chart of ozone.
Ozone Patterns
Conclusions
The distribution of ozone exhibits strong latitude dependence.
Total ozone amounts near the equator are rather low over the
course of each year, and increase as we move from tropics to
higher latitudes. This phenomenon can be explained by
stratospheric circulation, also known as Brewer-Dobson
circulation, which transports high ozone from the tropics to the
lower stratosphere of the high latitudes.
Buja et al. (1996) proposed a rudimentary taxonomy of interactive
view manipulations for high-dimensional data, and they are:
focusing individual views, linking multiple views, and arranging
many views.
The distribution of ozone also shows strong seasonality
dependence. Ozone amounts over the northern region are rather
high, with the highest amounts in April, May, and June, then
decreasing over summer. The lowest amounts are present in
October, rising again over the course of winter. In the winter, we
see ozone column amounts are large at high northern latitudes,
and low at the tropics. Moving into the summer, we find ozone
amounts at high northern latitudes falling off from winter time,
and at the same time, tropical ozone increases. Wind transport of
ozone is principally responsible for the seasonal evolution of these
higher latitude ozone patterns.
This web-based interactive data visualization demo system is one
example of focusing, linking, and arranging views; and it is also a
valuable complement to the techniques of exploration of large
multidimensional datasets which have spatial-temporal components.
REFERENCES
Buja, A., Cook, D., and Swayne, D. (1996) Interactive High-Dimensional Data
Visualization. Journal of Computational and Graphical Statistics, 5, 78–99.
Lefohn, A.S.; Simpson, J.; Knudsen, H.P.; Bhumralkar, C.; Logan, J.A. (1994 )
An evaluation of the kriging method to predict 7-h seasonal mean ozone
concentrations for estimating crop losses. Journal of Air Pollution Control
Association, 37, 595–602.