Scott T. Shipley, Ira A. Graffman, and David P. Beddoe

GIS Does the Weather

The Esri ArcInfo and ArcView Geographic Information Systems (GIS) are being used to manipulate, analyze and display hydrometeorological data and information. This paper addresses issues of GIS as a Weather Processing System (WPS), using a prototype GIS/WPS demonstration based upon the Esri GIS, and weather data from publicly available archives. This prototyping effort has demonstrated that the key functional components for weather systems are currently supported in a Commercial Off-The-Shelf (COTS) GIS. GIS and "traditional" WPS have been developed on parallel but independent paths, with WPS tracing its origins back to World War II. GIS has a similar heritage, with more than 25 years dedicated to the solution of "geographic" problems. High performance workstations, open systems, desktop computing, and the drive towards standards for coding and data interchange have brought the GIS and WPS benchmarks close to the break-even point. Just how close are they?


1. Development of GIS and WPS

GIS traces its origins to the 1970's with the advent of high performance workstations. The prime GIS problem has been to bring data and information from "uncoordinated" sources and on varied media into a common framework or projection so they can be superimposed, applying advanced graphic and video techniques to the age-old tradition of mapping and cartography. The Federal GIS community is now coming to grips with duplication of effort through the Federal Geographic Data Clearinghouse established by Executive Order 12096 (Clinton, 1994). The GIS industry is grappling with calls for Open GeoFunctionality Interoperability (e.g. Hecht, 1994).

At the same time and mostly independent of GIS, the meteorological community has been exploiting latest technologies to solve a different problem: the visualization of non-linear processes associated with atmospheric movements and changes in state. Early systems such as McIDAS made quick approximations for geography in order to support rapid animation of graphics on satellite images. The meteorological community currently supports a wide array of independently developed and mutually exclusive software systems, as evidenced by the Extended Abstracts for Interactive Information and Processing Systems (IIPS) (Shipley and Beddoe, 1996), now in its twelfth year. WPS development appears to focus on two key factors: 1) speed and throughput of near-real-time information, particularly under animation and now 3-D, and 2) data compaction with impacts to communication and storage. These factors have contributed to the current wide array of encoded meteorological data formats which are protected and sanctioned by International Agreements.

2. Weather Processing System Goals

Interesting discussions are underway concerning the roles and operations of emergency, public and commercial services which gather and disseminate weather information. Some argue that high performance workstations will allow WPS to support automated forecasting. Others maintain that the prime focus of WPS should be office automation. Current high-powered WPS applications such as "dataset browse during animation" demonstrate what is possible, but do not leave much room for other applications operating simultaneously. Disregarding such advanced applications, we suggest that the bulk of the work needed in a "weather support" situation falls into the office automation category. Consider weather support functions in this order of priority:

1) Maintain lists of contacts, observers, and equipment, their positions and contact methods,
2) Address and forward time-critical information to/from contact points (collect and disseminate information),
3) Quality control critical information (usually done last or on a time-available basis),
4) Compare current observations/forecasts to thresholds and climatology, and
5) Watch the weather.

Current COTS GIS technology provides the means to automate such weather support functions without major changes to existing data formats. To prove this point, a prototype GIS/WPS should demonstrate, at a minimum, the following functionality:

a) Display Meteorological Data, including station plots, model grids, radar and satellite images,
b) Use Intrinsic GIS Functionality to support traditional WPS functions, including contouring, cross sections, and superposition,
c) Incorporate User-Supplied Algorithms and software through calls to subroutines (e.g. C or C++) and clients,
d) Support an Interactive GUI Builder, and
e) Animation.

3. Results of Esri Prototype Experiment

Demonstrations of these GIS/WPS capabilities have been prototyped by Esri personnel. "Live" operations have been presented with real (archived) meteorological data using ArcInfo version 7.0.3 and ArcView 2.0 on an HP 735 UNIX platform with the Informix RDBMS. Efforts are underway to optimize the Esri GIS integration prototype for operational hydrometeorological environments.

3.1 Display Meteorological Data: GIS manages meteorological data as points, lines and polygons. Images (including raster datasets such as DEM) are usually referenced "as is" and are often displayed as backdrops to geographic overlays. Examples for point data include lightning strikes and surface station observations. Lines are collections of connected arcs with node points. Examples for line data include rivers and streams, with gage or river forecast point positions associated with selected nodes. Polygons are collections of topologically connected lines, including shared lines for adjacent polygons. A sample polygon approach to NEXRAD polar data is shown in Figure 1, where polygon shading is used to present radar reflectivity.

Figure 1 [28k gif]
Figure 1. Polar coordinate radar reflectivity (1 km) for the Dover, DE NEXRAD, plotted as adjacent polygons on a Lambert Conformal projection. Intrinsic GIS functions are also shown for user selection of county polygons and marking of the selected county records in an attached relational database (RDBMS).

3.2 Intrinsic GIS Functionality: ArcInfo currently provides over 3400 callable functions supporting a wide range of spatial database, analysis, display and user interaction operations. The contouring function is demonstrated in Figure 2, where intrinsic GIS macros were used to ingest, smooth and display surface station measurements onto a 60 km rectangular gridfield.

Figure 2 [11k gif]
Figure 2. Contours for wind speed [knots] computed at 60 km resolution over CONUS and Alaska based on surface station observations. Intrinsic GIS interpolation methods were employed, but calls to user supplied algorithms are also supported.

3.3 User-Supplied Algorithms: Significant coding productivity can be realized through use of the Esri ARC Macro Language (AML) and Avenue, an object oriented programming environment. Current GIS and related technologies are realizing lower costs and risks for implementation of applications compared to implementation in C or C++ (Strand, 1995; Zhuang and Engel, 1995). For example, the GIS/WPS prototype for a Pilot Briefing application shown in Figure 3 was accomplished in 440 LOC (including comments) using Esri AML.

Figure 3 [16k gif]
Click picture for full image, 1200x825 (84k).
Figure 3. A prototype application for the Pilot Briefing function. User marks flight plan waypoints on a shadefile (upper panel). Stations within 25 km of the flight path are selected and plotted at the surface of a vertical cross section using a Digital Elevation Model or DEM (lower panel). Weather data is available by station ID or symbol through an attached RDBMS.

3.4 Interactive GUI Builder: User programmable Visual Graphical User Interface (GUI) toolkits are standard COTS components in ArcInfo and ArcView.

3.5 Animation: Intrinsic prototype animation was too slow in comparison to current WPS technology, but can be provided using third party software. Is the GIS industry working on this capability?

4. What's Next?

GIS provides a powerful medium to integrate existing analytical and data resources. Drutman and Rauenzahn (1994) describe the integration of existing software and data into a Marine Geophysical Modeling System using ArcInfo as an enabling technology. Jones et al. (1995) discuss mapping the effects and meteorology of the Chernobyl accident, using ArcInfo to coordinate efforts across the European Union and the Republics of the Commonwealth of Independent States. EIS International (1995) recently announced the complete integration of ArcView GIS into its Emergency Information System (TM) as shown in Figure 4, where predictions of atmospheric dispersion are overlaid on a detailed map for real-time emergency support operations.

Figure 4 [65k gif]
Figure 4. Within seconds of an atmospheric release, an EIS/Win user predicts plume dispersion using the ALOHA model and locates the release point on an EIS/Map image. The model results are then plotted on a detailed ArcInfo map using ArcView 2.

4.1 http://geog.gmu.edu/gess/wxproject/: Concepts for manipulating weather data are demonstrated on the Internet using ArcView 1 by the GIS Weather Project at George Mason University. Sample weather data including the lightning and radar data referenced in this paper are currently provided at this location (subject to change), including all themes required to construct Figure 5. This figure demonstrates the use of ArcView GIS to merge meteorological data as points (lightning), lines (river levels) and polygons (radar reflectivity), correctly positioned over a satellite image.

Figure 5 [85k gif]
Figure 5. Weather Project sample window from ArcView 2, showing lightning, radar and geography on a georeferenced Landsat image (data samples from various dates). Intrinsic GIS function also shown for NWS office identification and marking from an attached relational database.

4.2 http://iwin.nws.noaa.gov/: Live weather data are now available to the public at no charge by Internet distribution, and radio broadcasts courtesy of the US Government. We have explored the experimental radio broadcast in the metropolitan Washington, DC area on VHF at 163.350 MHz, using ArcView to display surface station observations in near "real time". As shown in Figure 6, the Emergency Managers Weather Information Network (EMWIN) packet radio broadcast can be captured, decoded and adapted for tactical displays of meteorological data. The EMWIN broadcast includes all NWS Watches and Warnings, river gage reports, and other information of interest to existing and future users of GIS. At the time of this writing, the EMWIN broadcast is available in Norman and Tulsa, OK, as well as from a satellite transponder.

Figure 6 [16k gif]
Click picture for full image, 1069x706 (93k).
Figure 6. Emergency Managers' Weather Information Network (EMWIN) sample window from ArcView 2, showing live surface station reports displayed over the Susquehanna River Basin. Surface Airways Observations (SAOs) are decoded on the fly (sa2.dbf) and linked to a point event theme (Stations.dbf).

The GIS industry continues to innovate in areas currently unaddressed within the WPS community. WPS development efforts typically exploit new technologies as they become affordable. For example, Esri announced in 1995 the acquisition of a Spatial Database Engine (SDE) capable of supporting 100+ users with simultaneous access to very large and continuous spatial databases. Oracle and Esri subsequently announced technology which bundles SDE with the Oracle 7 Spatial Data Option. Using AML and Avenue, the authors have experienced applications coding productivity increases over traditional methods by factors of 5 to 10. Emerging methodologies including Open GeoData Interoperability (Tcl/Tk et al.) are therefore expected to overtake such traditional development methods on grounds for reduced cost and lower project risk.

ACKNOWLEDGMENTS

This abstract first appeared in the Extended Abstracts of the AMS 12th IIPS (Shipley and Beddoe, 1996), and has been updated. Prototype designs include contributions by Roger Shriver (NOAA SAO) and Mike Heathfield (NWS). Mark Smith and Elizabeth Falkenberg (Esri) produced the prototype in two weeks and at Esri expense. Grid and surface station observations courtesy of the Air Resources Laboratory (ARL). Lightning data courtesy of Global Atmospherics Inc. NEXRAD base reflectivity courtesy of Alden Electronics. Figure 4 courtesy of EIS International.

DISCLAIMER

The first two authors are currently on assignment to NOAA NWS, where they support the Advanced Weather Interactive Processing System (AWIPS). This paper does not represent official positions nor endorsement by NOAA, the National Weather Service, or the AWIPS Prime Contractor.

REFERENCES

Clinton, B., 1994: Executive Order 12096, Coordinating Geographic Data Acquisition and Access, Federal Register, 59, no. 71, 17671.

Drutman, C. and K.A. Rauenzahn (1994) Marine Geophysics Modeling with Geographic Information Systems, Oceans 94, 3.

EIS Intl., 1995: EIS/InfoBook and ArcView 2, New GIS Interface Delivers Greater Integration and Performance. Hazard Technology, XV, no. 1.

Hecht, L.G., 1994: Open GIS Foundation Fosters Industry Synchronization. GIS World, 7, no. 6, 24.

Jones, A., M.V. Liederkerke, M. De Cort, and A. Cooper, 1995: Mapping the Effects of Chernobyl. Joint Research Centre Environment Institute, Esri 1995 User Conference, Proceedings.

Shipley, S.T. and D.P. Beddoe, 1996: The ArcInfo GIS as a Weather Processing System. 12th International Conference on Interactive Information and Processing Systems, Atlanta, GA, American Meteorological Society, Boston, MA.

Strand, E.J., 1995: End User Programming of GIS Applications. GIS World, 8, no. 5, 40.

Zhuang, X. and B.A. Engel, 1995: Tcl/Tk GUI Toolkit Offers Cross-Platform Application Development. GIS World, 8, no. 7, 58.


Scott T. Shipley, Hughes STX Corporation
Affiliate Professor, Dept. Geography, MS 1E2
George Mason University, Fairfax, VA 22030-4444
sshipley@gmu.edu

Ira A. Graffman, Hughes STX Corporation
4400 Forbes Blvd, Lanham, MD 20706-4392
irgraffman@delphi.com

David P. Beddoe, Esri Federal Systems
2070 Chain Bridge Road, Vienna, VA 22182-2536
dbeddoe@Esri.com