St. Joe Corporation is a large paper company that has been actively acquiring land in Florida and Georgia for over 100 years to support their forestry operation. Currently, their holdings include over 1 million acres. In an attempt to track properties in a more systematic and accurate fashion, St. Joe began to express an interest in GIS technology. A parcel-based topologically intact polygon coverage was proposed to replace St. Joe's existing paper tax map collection. This trend was reinforced once Peter Rummell took on the role of CEO for St. Joe, and expressed an interest in pursuing an aggressive real estate initiative.
In early 1997 Michael Baker Jr. Inc. entered an agreement to provide
GIS consulting services to St. Joe Corporation. The proposed scope
of work described robust and innovative real estate GIS for St. Joe. Initially
Baker GIS was hired to perform a host of activities including:
Develop a detailed Hardware and Software Specification
A GIS server and workstation was purchased, as well as a high resolution color plotter, a laser printer, 2 support/client workstations and a large screen display for presentations to management. This was all networked together via a 12 port 10/100 auto negotiating Ethernet switch. These were envisioned to be the core devices for the Jacksonville segment focused on GIS processing, production and visualization. All of these systems were purchased, installed, appropriately configured and bench marked to determine effective throughput and to see their impact (if any) on the greater St. Joe network.
An Esri solution was selected for St. Joe because of the high quality spatial tools that Esri brings to the user. The primary GIS software purchase included a Workstation ArcInfo 7.1.1 for Windows NT installation, ArcView 3.0a as the primary visualization and output client, and ArcPress for ArcView as a reliable output creation tool. ArcInfo was installed on the server, and ArcView was put on the workstation. A small amount of initial customization of ArcView was undertaken to bring the most commonly used tools, scripts and extensions available to the user/analyst. ArcExplorer, or a compatible custom developed viewer client, was envisioned to be the "thin-client" GIS viewer/query tool.
A three tiered implementation was envisioned for deployment for the
GIS, allowing map maintenence and spatial processing to be done at the
first level, visualization and report production on the second, and rudimentary
viewing, querying and printing at the third (Figure 1.) In this way
St. Joe could reconfigure each modular component of the GIS to better meet
the needs of each user group. Customization could be compartmentalized,
allowing different departments within St. Joe to influence the look and
feel of the tools that they use, and allowing different contractors to
focus their efforts on specific projects, thereby exploiting techical know
how based on industry specialization and experience.
Figure 1. A three tiered approach to GIS implementation
St. Joe's Forestry GIS office in Port St. Joe on the Florida Panhandle was linked to Jacksonville via a PSTN T1 connection. This facilitated sharing of data between the GIS server in Jacksonville and tree stand data (and other coverages) located in remote Port St. Joe. Remote access for clients and contractors was provided via a remote access server in conjunction with specialized remote control software. This software, when used over a dial-up (POTS) connection facilitated managing the system, configuring the server, and text editing such as AML coding, but was too slow for mapping purposes. A remote ArcView session, even when constrained to 256 colors at VGA resolution could take minutes to refresh over the dial-up connection. Also, the dial up connection was expensive for St. Joe, and consequently of limited use from a map making and data processing perspective.
The creation of a virtual private network or "extranet" was proposed
to link all remote vendors to the GIS data server, and provide high speed
data access/processing (Figure 2). Negotiations were begun with St. Joe.
Corporation's IT department to create a new subnet to function as a "DMZ"
for the GIS network segment. In conjunction with an implementation of Microsoft's
Point to Point Tunneling Protocol (PPTP), this DMZ would effectively isolate
non-authenticated traffic from the greater St. Joe domain while leveraging
the installed T1 access to the Internet. Once the GIS server was configured
to support tunneling each contractor would be able to access data and applications
on the server, manage the system, and administer the database. While this
was ultimately postponed in favor of dial-up access, further development
of the GIS will necessitate greater bandwidth.
Figure 2. Proposed IP tunneling/VPN approach towards high-speed remote access.
Evaluate Existing and Available Data Sources
Over the course of several months St. Joe had acquired a great deal of digital spatial information from various municipalities, county mapping agencies and data vendors. This data was of various sources, dates, scales, projections and coordinate systems, thus it proved much more valuable for smaller focused projects as opposed to the greater GIS. In the course of gathering this data, a metadata store had been compiled. It included large notebooks of information, as well as a MS Access database containing file naming conventions, source information, and some projection data. Baker was charged with reviewing this data, categorizing it in terms of overall usefulness, and supplementing it with a variety of other data from vendors and government agencies. Some data was also created via digitizing and spatial processing/overlay analysis.
Perhaps the most valuable of all of the data sources was determined to be the Florida Department of Environmental Protection's (FDEP) GIS data repository, which can be easily accessed via the World Wide Web. FDEP data, available for free download, contains a variety of consistent, precise and useable data from various Florida water management districts and other agencies. The data is stored in e00 Arc export format, which facilitates rapid processing and inclusion in the GIS. Ultimately FDEP data, already extensively used by Forestry, was to become a major resource for the GIS, and directly influence the structure of the GIS parcel database. Some other data sources included TIGER 94, DLG, DEM, DOQQ, LandSAT TM, IRS-1C, landcover data from Florida Department of Wildlife and Freshwater Fisheries, and various data vendor products including scanned USGS topographic maps. All of this data was populated onto the GIS server, as well as other mirrored servers at Baker facilities for the process of making maps.
The following subset of themes were selected to represent the primary
GIS database for St. Joe corporation:
Attribute/Tabular Data Structure
A prototype parcel mapping database was designed by Michael Baker Jr. Inc. to allow St. Joe to track it's various non-contiguous land holdings in relation to several themes deemed important from a real-estate perspective (Figure 3.). The primary application of the database was to allow St. Joe management access to pertinent corporate information. Some of these attributes included an assessor's database, tax information on the parcel level, ownership (for parcels directly adjacent to St. Joe holdings), landuse, some COGO data, demographic information (Census), wetlands and wildlife (endangered species) information, to name a few. The database was designed with certain functionality in mind, such as the ability to track assessed value and taxation amount over time, and the ability to determine the percentages of various landuse categories and wetlands that make up each parcel, and the ability to track imagery of diverse projections, resolutions and dates. Situations requiring many-to-many relates were identified and built into the design. Also, an effort was made to offload potentially sensitive information, such as the assessor's and ownership tables, onto a third party RDBMS such as Oracle. This would allow free data viewers such as Esri's ArcExplorer to view pertinent data, such as a parcel's location, throughout St. Joe's Corporate Data Network, but without certain attributes. Only approved viewers capable or establishing an SQL Connect would be given access to all related tables.
Figure 3. Parcel based database design to support St. Joe's
property mapping applications
Coordinate Geometry Considerations
Baker recommended that St. Joe "unproject" all vector coverages and store them as double precision in a Librarian/ArcStorm transaction based environment. The reasons for this were several.
Ultimately a system was designed that would support and manage a parcel database, allow users to contrast property holdings with other spatial data sets deemed relevant from a real-estate development perspective, and make attractive maps. This system would reference a meta-data database describing projection parameters for all acquired imagery and automate the ArcView view-document window dynamically. In doing so the system would effectively isolate the user from having to chose projection information, allowing a much more simplified approach towards map creation through automation.
A prototype parcel mapping application was coded in Avenue (Figure 4.)
to support the display, query and overlay analysis of St. Joe properties
at the parcel level. This parcel browser allowed users to browse through
the St. Joe parcel database, turn on and off layers, run pre-coded analytic
queries based on user variables, and make attractive maps. A "St. Joe ArcView
Extension" was proposed to facilitate consistent and deployable customization
throughout St. Joe Corporation. This extension would include the Parcel
Browser, An analytic overlay and statistical analysis tool, and a custom
ArcView palette to support St. Joe map symbology. A prototype interface
was designed and presented to St. Joe to review, and skeleton code was
generated to support some initial functionality. St. Joe Forestry's GIS
office was consulted regarding the potential development of forestry applications,
and several types of functionality were discussed. Ultimately the St. Joe
AV Extension was envisioned to be a collaborative effort between Baker,
and several other contractors. Each party was to add some level of organization,
coding functionality, data model design or standardized symbology to the
extension and it's components. It was hoped that a collaborative approach
would drive a better product for St. Joe, and that each participant would
take a bit of personal pride in the overall functionality.
Figure 4. St. Joe Parcel Browser via customized ArcView Extension
An ArcView extension was proposed in lieu of other customization options for several reasons:
A process was outlined using batch processing in AML to import and build the FDEP export files, once downloaded onto the GIS server. It was envisioned that the map sheet tiles would then be converted to double precision, undprojected, and re-built into map sheet tiles per appropriate feature class. Once pre-processed and checked, a secondary procedure was developed to populate additional items added to the FAT, including several items to be related to lookup tables ant other data sources. Finally, once re-checked for appropriate domains and ranges, these tiles were to be integrated into the GIS proper, and made available for mapping purposes. Certain spatial operations were envisioned to be a part of the conversion process, such as the polygon overlay of parcels with several feature classes, to produce new coverages of complex polygons or regions. These new features where several identical PINs were associated with other features were to provide a proportional index of each specific feature class for each tax parcel. This information was to be used later during analytical processing to create selection sets based on site-specific mitigation factors, and user/analyst preference.
GIS has the potential to facilitate an exciting, broad based restructuring of the way that St. Joe manages it's land holdings. A spatial database would enable St. Joe and it's affiliates to re-think it's land management, development, and acquisition goals from a rule driven perspective. The process of spatially enabling and centralizing St. Joe's corporate databases could lead to improved efficiency, access to better, more accurate corporate information, and ultimately larger profits.