EPA/600/A-97/018
Visualization of Remotely Sensed Data with 3-D Meteorological Modeling Results
YanChing Q. Zhang and Jeff Wang
Lockheed Martin/US EPA Scientific Visualization Center
Sharon LeDuc and Jonathan Pleim
National Oceanic and Atmospheric Administration
on assignment to Atmospheric Modeling Division, NERL, US EPA
Research Triangle Park, NC 27711
ABSTRACT
EPA Models-3 system, the third-generation air quality modeling system, is a framework which
accommodates new science modules and support their evaluation. One component of the Models-3
system is the Mesoscale Meteorological Model (MM5) which simulates clouds, water-vapor, wind and
air deposition. Without comparing simulated data to measurements of real atmospheric data, the quality
of the simulated data is completely unknown. There are tremendous amounts of data from satellite and
radar which after being combined with geographical information may be compared to the simulated data.
Visualization provides a quick and easy tool to present and analyze remote sensing data sets and
compare with the modeling results.
Our purpose in this paper is to investigate how we can determine the quality of simulations from
MM5 using available software tools to display both simulated and measured data. This capability will
help environmental scientists to modify meteorological simulations in the MM5. We will compare the
satellite data with model predictions by overlaying the remote sensing satellite imagery with the
modeling data using two software tools, the McIDAS and Vis5D. Output from two variations of the
MM5 are visualized along with the remotely sensed data imagery from Geostationary Operational
Environmental Satellite (GOES).
1. Introduction
1.1 Overview
Models-3, EPA's third-generation air quality modeling system, is being designed to be user
friendly, to overcome known weaknesses in existing modeling systems and to accommodate new
science models and evaluation of all components of the modeling systems (Dennis et al., 1996).
A meteorological model is a necessary component of Models-3. There are several
meteorological models widely used in the environmental community to simulate the required
meteorological variables. The first one of these models to be included in the Models-3 system is
the Fifth Generation NCAR/Penn State Mesoscale Meteorological Model MM5. Clouds, water-
vapor, and wind are some of the variables simulated and output from this model.
One of the critical problems facing environmental scientists is determining how well the model
reflects reality. Without proper evaluation, modeling results are open to question. Spatially and
temporally dense data are needed for such an evaluation. Fortunately, there is an increasing
amount of data being collected from satellite and radar sources which can be combined with the
geographical information so that the registration is accurate. Computer visualization enables
scientists to look at the model output and to compare effectively with the remote sensing data.

-------
1.2 Project Background
Clouds play a major role in air quality. Cloud properties are critical in the chemical
transformations in the atmosphere and precipitation from clouds removes pollutants from the air.
Aqueous reactions in clouds differ from gas phase reactions. MM5 simulates various effects of
cloud processes and the cloud modeling schemes are crucial to a realistic simulation of
mesoscale meteorology and air quality. Therefore, evaluation of MM5 predictions through
comparison with satellite cloud imagery is very valuable (LeDuc et al, 1996). Scientific
visualization can be used to compare meteorological model results with the satellite imagery on
the desktop computer, provides greater insight or understanding of how well the models are
approximating reality.
This paper and the related research is a collaboration between US EPA, Lockheed Martin (EPA
SVC), and the Earth Science Laboratory of the University of Alabama at Huntsville. It is funded
by EPA under the third generation air quality modeling system, Models-3 project. The
visualization of the satellite data with the meteorological results are conducted by the EPA
Scientific Visualization Center (SVC). The SVC is responsible for the examination of the
scientific visualization techniques and tools. Specifically, visible and infrared images obtained
from the GOES satellite are compared with MM5 modeling results.
This paper describes how we examined the clouds simulated by the MM5 model using available
software tools. This examination determined when simulations were in error and thus provides
help to environmental scientists who want to improve the MM5 cloud parameterization. This
paper also describes how to compare satellite data with model predictions, by overlaying the
remote sensing satellite imagery with the 3D meteorological model MM5 output. The satellite
data used are from GOES satellite imagery obtained from the GOES Pathfinder Operation at the
University of Wisconsin; and the meteorological modeling results are from MM5, The software
chosen for the task includes Vis5D and Man computer Interactive Data Access System
(McIDAS). Finally, the MM5 cloud-related information is visualized simultaneously with the
GOES remotely sensed imagery.
2. Visualization of the remotely sensed data
2.1 Available Software:
2.1.1	ESR I®ARC/INFO®
ARC/INFO is a leading Geographical Information System (GIS) software marketed by
Environmental System Research Institute, Inc. (ESRI). It is designed to process spatial and
statistical geographical information datasets. ARC/INFO also provides some limited capabilities
of handling remotely sensed data. The software is more widely used by GIS professionals.
2.1.2	PCI® EASI/PACE®
A commercial package from PCI Remote Sensing Corp.'s remote sensing software family,
EASI/PACE, offers a wide range of functionality. EASI/PACE provides extensive capabilities in
image classification, geometric correction, orthorectification, enhancement, filtering, vector
editing with image backdrop, terrain analysis and visualization, radar image processing, DEM
(USGS Digital Elevation Model) extraction, atmospheric correction, and hyperspectral data
analysis. EASI/PACE is a modular system. The Image Processing Kit is the core of EASI/PACE,

-------
the turnkey solution for the PCI remote sensing software family. Depending on processing
requirements, additional software packages may be added.
2.1.3	ENVI®
Environment for Visualizing Images (ENVI) is a robust, easy-to-use image processing system.
ENVI provides analysis and visualization of single-band, multispectral, hyperspectral, and radar
remote sensing data. ENVI's point-and-click graphical user interface (GUI) makes it easier and
faster to learn and utilize the system to process images. ENVI is built on Interactive Data
Language (IDL), a structured, array-based programming language that includes flexible image
analysis tools.
ENVI's flexible data handling allows most data files to be read directly off the disk (including
CD-ROM) without converting them to another format. ENVI uses a generalized raster data
format that consists of a simple flat binary file and a small associated ASCII (text) header file.
This approach allows reading of almost any image format into ENVI, including those with
embedded header information. The ASCII header file provides ENVI with information about the
dimensions of the image.
ENVI supports both image-to-image and image-to-map registration through Ground Control
Points (GCPs).
2.1.4	ERDAS® IMAGINE®
Another commercial product is available from ERDAS, Inc. It is called IMAGINE and offers
total imaging GIS solutions on UNIX-based workstations as well as on PCs. ERDAS IMAGINE
is compatible with ARC/INFO. ERDAS IMAGINE actually has ESRI's ARC data model built
in.
2.1.5	Lockheed Martin MeteoStar® LEADS
Lockheed Environmental Analysis and Display System (LEADS) is a system for the receipt,
integration, and processing of all types of meteorological data, and interactive generation and
display of tailored weather support products. It is specifically designed as a meteorological
application.
2.1.6	McIDAS and Vis5D
McIDAS (Man computer Interactive Data Access System), under development since 1970 at the
Space Science and Engineering Center (SSEC) of the University of Wisconsin-Madison, is a
sophisticated, video interactive set of tools for acquiring, managing, analyzing, displaying, and
integrating data.
McIDAS generates multicolor composites of conventional and satellite weather data in a variety
of displays in two and three dimensions as well as time-lapse sequences of these analyses.
Designed to handle large amounts of meteorological imagery and other atmospheric data in a
convenient manner, the system is a vast resource of image-processing and applications programs
and subroutines. McIDAS hardware and software are used throughout the world.
Vis5D is a system for the interactive visualization of large 5D gridded data sets (3D in space
with time sequence and species' concentration) such as those made by numeric weather models.
VisSD was written by the Visualization Project at the University of Wisconsin-Madison Space

-------
Science and Engineering Center (SSEC) by Bill Hibbard and Brian Paul. One can make
isosurfaees, contour line slices, colored slices, volume renderings, etc. of data in a 3D grid, then
rotate and animate the image in real time. There's also a feature for wind/trajectory tracing, a
way to make text annotations for publications, etc.
Vis5D provides superior 3D visualization capability and is free, but does not provide internal
geopositioning functionality. MclDAS is used for the geopositioning because it is a standard
tool used with meteorological data, including satellite data. Considering the labor, availability of
local expertice, and cost of the software, Vis5D and MclDAS was determined to provide the
most cost effective solution for us. The flexibility of the software allows us to visualize 3D
meteorological modeling results and georeference the satellite map with the model results
without writing internal code.
2.2. Methods
The GOES satellite data were obtained from the SSEC, Wisconsin State University GOES
Pathfinder Operations. MclDAS is used to pre-process the satellite (GOES) imagery. First, using
MclDAS, GOES imagery is projected from native projection to Lambert coordinates. Second,
the imagery is subsected to the domain of the model. Finally, VisSD is used to visualize the
meteorological modeling results with the GOES imagery.
Figure 1 shows two satellite images before and after the pre-process from MclDAS. Figure la is
the original imagery from GOES Pathfinder Operations at the University of Wisconsin. This
image is in the original satellite projection (Mercator). Figure lb is the processed satellite image
in Lambert Conformal projection, which is subsected to a domain size identical to the
meteorological model (MM5).
VisSD is used here for the visualization of the output of the NCAR/Penn State Mesoscale
Meteorological Model (MM5). Figure 2 shows the rendering of the isosurfaees of the cloud
water (the blue volume) and the rainwater (the white yellow volume). They are overlaid with the
corresponding satellite images and other information. The United States map is superimposed on
the top of the satellite imagery.
3. Results and Discussions
3.1 Comparing Model Prediction with Satellite Data
Shown in Fig 3 is the GOES satellite imagery infrared band overlaid with the MM5 results. The
period modeled is on 08/02/88. The MM5 modeling results shown are at a 54 km grid size.
Clouds and rainwater are visualized as isosurfaees and normalized by humidity, the interiors of
which have a normalized water contents greater than 0.01.
The infrared information from GOES indicates temperature, which can be compared to model-
predicted temperature wherever clouds are present. This is accomplished by projecting the
model-simulated temperature onto the isosurface of the cloud water. In the infrared band, clouds
that are highest are also the coldest. Therefore, when the infrared band is projected onto the
Earth's surface, the highest clouds appear the whitest. The model's predicted isosurfaees are
translucent so the infrared cloud projections can be viewed on the surface. White areas on the
surface are clouds detected by infrared sensors on GOES; they correspond to the model-predicted
clouds with coldest temperatures.

-------
3.2	Sidc-by-Side Comparison of the Two Model Runs
Two MM5 model outputs with different grid resolutions (80km and 54km) are compared side-
by-side in Figure 4. The grid resolution is not the only difference as the internal equations have
different assumptions regarding the hydrostatics. This allows examination of cases in which one
model configuration (resolution and hydrostatics) does a better simulation than another model
configuration based on the comparison to satellite imagery. Observed satellite data offer good
spatial and temporal resolution and provide the opportunity to gain additional insight in inter-
model comparisons.
3.3	Velocity Fields Predicted by Model
Vertical wind information is displayed in Figure 5 and shows the detailed physics of the model
and the structure of convective events. Note the shafts with upward motion, associated with large
clouds. High, cold cloud tops should exist in these convective areas. The infrared satellite
information is added for comparison, with white representing the coldest clouds.
4.	Summary
In this paper, we described how we can examine certain output from a meteorological model
using software tools and thus help environmental scientists to improve their meteorological
models. Satellite data were from GOES and were obtained from the GOES Pathfinder Operations
at the University of Wisconsin. The meteorological modeling data were from MM5. We
described how we compared the satellite data with model predictions by overlaying the 3D
scientific modeling data over the remote sensing satellite imagery using software tools, McIDAS
and VisSD. Other software were discussed, but were not used in the comparison. The technical
issue, the geoposition of the data with the maps, the cost, the local expertise, and the labor were
the reasons we chose the approach using McIDAS and VisSD.
Finally, three-dimensional output simulated in MM5, including cloud water, relative humidity
and velocity, were visualized simultaneously with the remotely sensed data imagery (GOES).
Observed satellite data provide the opportunity to gain additional insight in inter-model
comparisons. Satellite-observed clouds offer good spatial and temporal resolution for this kind of
comparison. Through an improved understanding gained in using visualization tools, satellite
data can be incorporated into model predictions. Improved analytic tools for directly assessing
the accuracy of spatial predictions offer additional challenges and opportunities.
5.	Acknowledgment
The work performed by YanChing Q. Zhang and Jeff Wang was supported by EPA contract 68-
W2-0025 with Lockheed Martin.
6.	Disclaimer
This paper has been reviewed in accordance with the U.S. Environmental Protection Agency's
peer and administrative review policies and approved for presentation and publication. Mention
of trade names or commercial products does not constitute endorsement or recommendation for

-------
7. References
ARC/INFO:	http://www.esri.com/
Dennis, R.L., D.W. Byun, J.H. Novak, K.J. Gailuppi, C.J. Coats and M.A. Vouk (1996): The
Next Generation of Integrated Air Quality Modeling: EPA's Models-3, Atmospheric
Environment, 30, nl2, 1925-1938.
ENVI:	http://www.envi-sw.eom/
ERDAS IMAGINE: http://www.erdas.com/
LeDuc S., J. Pleim, Y.Q. Zhang and J. Wang (1996):GOES Visible and Infrared Channels to
Evaluate Cloud Processes in Air Quality Models, Second International Symp. and Spatial
Accuracy Assessment in Natural Resources and Environmental Sciences, May 21-23,
1996, Fort Collins, CO, USA.
McIDAS:	http://www.ssec.wisc.edu/software/mcidas.html
PCI EASI/PACE: http://www.pci.on.ca
Vis5D:
http://www.ssec.wisc.edu/~billh/vis5d.html

-------
Figure la: GOES satellite image before the "pre-process" using McIDAS software, in Mercator
projection.
Figure lb: GOES satellite image after the "pre-process" using McIDAS software, in Lambert Conformal
projection.

-------

Figure 2: Rendering image of the isosurfaces of the cloud water. They are overlaid with corresponding
satellite image.
Figure 3: GOES satellite imagery infrared band overlaid with the MM5 modeling results. The
temperature can be mapped (in color) onto the cloud isosurface.

-------

Figure 4: Side-by-side comparison of two MM5 model outputs, 4a: 80km grid resolution; 4b: 54 km
grid resolution..

-------
Figure 5: Vertical wind shows the detailed physics of the model and the structure of convective events.

-------
TECHNICAL REPORT DATA
1. REPORT NO.
EPA/600/A-97/018
2 .
3
4. TITLE AND SUBTITLE
Visualization of Remotely Sensed Data with 3-D
Meteorological Modeling Results
5.REPORT DATE
6.PERFORMING ORGANIZATION CODE
7. AUTHOR (S)
YanChing Q. Zhang1, Jeff Wang1, Sharon LeDuc2 and Jon
Pleim2
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
1	Lockheed Martin/US EPA Scientific Visualization
Center, Research Triangle Park, NC 27711
2	Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
10.PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
NATIONAL EXPOSURE RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
13.TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The EPA Models-3 system, a third-generation air quality modeling system, is a
framework which accommodates new science modules and supports their evaluation. One
component of the Models-3 system is the mesoscale meteorological model (MM5) which
simulates clouds, water-vapor, wind and air deposition. Without comparing simulated data
to measurements of real atmospheric data, the quality of the simulated data is completely
unknown. There are tremendous amounts of data from satellite and radar which, after_
being combined with geographical information, may be compared to the simulated data. •
Computer visualization provides a quick and easy tool to present and analyze remote
sensing data sets and compare with the modeling results.
Our purpose in this paper is to investigate how we can determine the quality of
simulations from MM5 using available software tools to display both simulated and
measured data. This capability will help environmental scientists to modify
meteorological simulations in the MM5. We will compare the modeling data using two
software tools, the McIDAS and Vis5D. Output from two variations of the MM5 are
visualized along with the remotely sensed data imagery from a Geostationary Operational
Environmental Satellite (GOES).
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/ OPEN ENDED TERMS
c.COSATI



18. DISTRIBUTION STATEMENT
19. SECURITY CLASS (This
Report)
21.NO. OF PAGES
JO
20. SECURITY CLASS (This Page)
22. PRICE

-------