Geographic Information Systems and E-Government Services

Jon Gant Assistant Professor Indiana University School of Public and Environmental Affairs Bloomington, IN 47405 jgant@indiana.edu (812) 855-0732

Wilpen Gorr Professor Carnegie Mellon University H. John Heinz III School of Public Policy and Management Pittsburgh, PA 15213 wg0g@andrew.cmu.edu

Danny Fernandes Ph.D. Candidate Carnegie Mellon University H. John Heinz III School of Public Policy and Management Pittsburgh, PA 15213 danny@cs.cmu.edu

This paper is a submission to the Workshop on Foundations of Electronic Government in America's Cities: A Multi-Disciplinary Workshop, March 8 and 9, 2001.

I. Introduction

The focus of this paper is to identify research needs in the area of E-government that combine geographic information systems (GIS) and advanced operations research, artificial intelligence, and statistical models. GIS greatly enhances public decision-making and government service delivery by making it easier to store, analyze and display land-related data. Because nearly 80 percent of government work involves information that is spatially organized, there is increasing demand for more effective use, management and development of GIS tools. Coupling this demand for better GIS tools has been a number remarkable advances in GIS technology that is making it possible to create innovative advanced GIS applications. Broadly speaking, we define advanced GIS as those applications that exploit decision-making or artificial intelligence tools, advanced database and mapbase technologies, and Internet and other advance communication protocols.

We present three cases that exemplify and motivate this research need and also use these cases to examine the managerial, technical and policy related barriers limiting the broader adoption of geographic information system (GIS) applications for E-government. These cases include: the Allegheny County Home Delivered Meals management and policy support system; the Park Grove Reservation System prototype; and, Snow Fighter, a GIS application used by the City of Indianapolis to manage snow removal. While these three examples greatly improve service delivery performance and enhance public decision-making, we raise the issue that advance GIS applications will be more broadly deployed in organizations that are better adept at dealing with the managerial, technical, and policy issues related to using GIS.

II. GIS and E-government 

A geographic information system (GIS) is a computer-based tool for mapping and analyzing things that exist and events that happen on earth. GIS technology combines geographic data and other types of data to generate visual maps and reports. GIS integrates common database operations such as query and statistical analysis with the unique visualization and geographic analysis benefits offered by maps. These abilities distinguish GIS from other information systems and make it valuable to a wide range of public and private enterprises for explaining events, predicting outcomes, and planning strategies . GIS is used in a broad range of public sector applications including, for example, land use and urban growth planning, legislative redistricting, crime tracking and law enforcement, benchmarking human services, emergency management, environmental monitoring, and public information services. 

While GIS is used broadly for descriptive tasks, recent significant technological advancements have enhanced the ability to create innovative advanced GIS applications. Current and emerging GIS applications now have the capability to integrate easy to use software and hardware technologies that allow users to perform such complex tasks as modeling patterns and trends, forecasting the impact of planning, policy, or strategy initiatives, and streamlining internal business and operation processes. Many of these advanced GIS applications go far beyond basic map-making. We posit that the most effective GIS applications for E-government will be ones that also include such capabilities as statistical analysis, operation research modeling and automated spatial modeling, incorporating real time or near real time data (e.g. global position systems data), database management (especially integrating enterprise-wide databases), user-controlled customization of analytical and modeling tools, Internet integration, and deploying expert systems and artificial intelligence. 

These advanced capabilities greatly enhance the potential to use GIS as tool for making management and policy decisions and for being a powerful tool for E-government endeavors. These capabilities should greatly improve the delivery of government services because they, among other factors, will:

The following three cases will illustrate the potential for GIS in enhancing E-government, particularly for improving service delivery. The first case describes a GIS application designed to support the Home Delivered Meals program in Allegheny County, Pennsylvania. This innovative application features a decision support modeling module, web integration, and interorganizational database integration, which are all necessary for improving this service in the greater Pittsburgh area. The second case is a prototype of a recreation area reservation system for the Allegheny County Department of Parks and Recreation. This system allows users to reserve park facilities over the Internet and is very innovative in how it uses an expert system to enhance this service. The third case describes Snow Fighter, which is a GIS application used by the Department of Public Works in Indianapolis, Indiana, to manage snow removal during winter emergencies.

Home-Delivered Meals System

A major goal of elderly persons is to remain in their homes and to avoid assisted living, nursing homes, and other residential institutions. Home-delivered meals, or meals on wheels, is a wonderful program that enables housebound elders to meet this goal. For example, Allegheny County, Pennsylvania has a population of 1.2 million and elderly population of about 200,000. Approximately two percent of the elderly use meals on wheels, if the service is available in their vicinity, yielding an estimated net savings of $40 million per year in avoided residential institution costs. There are 63 kitchens in the county, with an average of 65 or more clients each and four routes driven daily by volunteers to deliver hot meals. 

A problem with the network of kitchens is that it evolved over time, without central planning. Furthermore, there has been a migration of elderly populations out of city centers to suburbs (e.g., white flight in the 1960s and 1970s resulted in suburbanites who have aged in place and contribute to the migration). The network has not kept up with this population change. There are both obvious overlaps in kitchens that compete for clients within overlapping service areas, and obvious gaps in coverage of the elderly population by kitchen service areas. Our GIS and models show that there are also many non-obvious gaps and overlaps. 

We were funded by Pittsburgh's Jewish Healthcare Foundation to build a decision support system (DSS), employing GIS and operations research methods, to facilitate objective planning for meals on wheels kitchens. See Gorr, Johnson, and Roehrig (2000) and Gant, Gorr, et. al. (1997). We assembled an appropriate GIS using commercial sources for street network, zip code, block group, and municipal boundaries, and current year and five-year ahead projections of elderly population by block group. We did a census of the 63 kitchens, did some field work to estimate service times at stops, assembled needed data, and address matched the kitchens. 

At first we attempted simple solutions, such as placing circular buffers around kitchen locations. That obviously was too inaccurate because of the many factors affecting the size and shape of kitchen service areas; for example, the density and location of elderly population, density and quality of the street network, physical barriers such as rivers and limited access highways, the number of routes per kitchen and kitchen capacity in meals, etc. An additional factor is that kitchens plan on a maximum of 45 minutes for delivery of the last meal on a route to keep meals hot and volunteer drivers from being overtaxed. 

The solution approach, we found, requires a micro-level simulation of demand plus an interesting application of the so-called traveling salesman problem (TSP) algorithm. The objective was to estimate the service area of a single kitchen of specified capacities (number of routes and meals), placed on the street network. This would allow us to incrementally plug gaps and study overlaps in an interactive DSS. 

We collected and address-matched client stops data for a sample of 25 kitchens. We digitized a polygon around the stops locations to represent the service area of a kitchen. We then estimated the fraction of elderly using meals on wheels by dividing the number of clients by the estimated elderly population of the service area polygon (estimated using block group data and apportionment to service area polygons by area). Furthermore, we used a regression model to correct for an under-sampling bias of the resulting utilization rates, with an end estimate of two percent. 

The next step was to simulate stops for all block groups in Allegheny County using the utilization rate, block group elderly population forecasts, random point generation, and reverse geocoding to place randomly selected points on the street network. Random draws from a separate distribution estimated the number of clients per stop. The end result at this stage was a simulated point demand of stops and customers. 

The combined TSP and GIS solution is interesting. We divide the area surrounding a kitchen into equal-area pie wedges, with one route per wedge. Initially, each route has an equal share of the kitchen's capacity in terms of meals. We start with a small radius defining the outer edge of a wedge and apply the TSP to find an optimal route through stops within the wedge, returning to the kitchen. We keep increasing the radius and rerunning the TSP until a driver times out at 45 minutes or runs out of meals. If a driver times out and has undelivered meals, those meals are reallocated to another driver who has more delivery time, if available. 

Also included in the DSS is the ability to digitize barriers that represent another kitchen's turf or other limitations. We applied the DSS to optimally reallocate all existing kitchen capacity. The end result was maximally covered elderly population and gaps that we were confident were real gaps. We found that approximately 83 percent of the elderly population was covered with access to meals on wheels, while 17 percent was not. Some of the gaps identified were not obvious at first, but represent densely populated areas that "fell between the cracks." 

It is possible the place the meals on wheels DSS on the Web, along with estimated service areas of existing and planned kitchens. Then clients could identify their kitchens. Also, community groups could see where the gaps are and plan to fill them, including estimating the area that could be served. Such an activity would be done, naturally, one new kitchen at a time by individual groups (like a church). Hence, the particular DSS that was built is ideal for this use. 

In summary, this was a challenging problem. The only solution that we could find was a quite sophisticated GIS-based DSS.

Park Grove Locater and Registration System

Allegheny County, Pennsylvania has two large parks, North Park and South Park, with many groves each that can be reserved for picnics and other social gatherings. Each grove has barbeque cooking facilities, picnic benches, and various recreational facilities (e.g., swings, tennis courts, horse shoe courts, etc.). 

At present, to register for a grove, a citizen must travel to the parks and recreation department, identify a grove with an available time slot that meets capacity and use needs, and pay a use fee. This is an ideal application for e-government, GIS, and some rule-based expertise in identifying groves for use. This is a system that county authorities would like to implement in the near future. 

The GIS would have the following layers. First, zoomed out, would be the entire county, major roads, and the parks. Zoomed in would be particular groves with facilities digitized and annotated. Also available would be a tabular listing of capacity, particular facilities, etc. and digital photos. Out-of-town guests could get A to B routing to a grove from major routes. 

The citizen desiring to reserve a grove would enter fill out a form to specify needs; for example, date, time interval, number of guests, recreational facilities needed, etc. Then a rule base would score available groves and allow the citizen to review top matches, using the GIS and various supporting materials. On making a choice, the citizen could register, make payment, and complete all arrangements. Later, after use, the citizen could be emailed and asked to fill out a survey on satisfaction, suggestions, etc. The results of surveys could be included as supporting materials for site selection (and improvement by parks and recreation). 

In summary, this example has the desirable one-stop feature of e-government, combined with several GIS applications to make accurate selection feasible and fast. Key is the smart query system, incorporating some expert rules and fuzziness, to quickly provide valuable information.

Snow Fighter

Snow removal is a critical local government service that can be very costly to manage. In general, managing snow removal is a slow process due to operational burdens, namely the time it takes to clear the snow, and, because of the administrative burdens. Obviously, because of the unpredictable nature of winter storms from year to year, managing snow removal is a great challenge for most local governments. Local decision-makers are often uncertain about the timing of winter storms and snow fall amounts, making it difficult for public works managers to determine the best snow removal strategy and how much equipment, material and people is needed. Finally, if the area affected by the storm is declared an emergency officially, local governments apply to federal and state emergency management agencies once the snow is removed to recover portions of the money spent on snow removal. 

Snow removal in Indianapolis, Indiana is very similar to most other cities. The city is divided into four districts based on the current location of three garages. The garage are the staging locations for snow removal trucks and serves as a place where the drivers get their snow removal instructions, and where the trucks are loaded down with salt and other traction management materials. There are 611 snow routes that were created over a decade ago. Each route is assigned one of three priority levels including, priority snow routes that are major streets and thoroughfares, priority residential routes, and all other streets. Key information that is collected and tracked during the snow removal process includes information about the drivers, contractors, trucks, equipment status, number of actual hours worked, amount of salt and chemicals used, etc. Additionally, managers consult with weather experts, track the number of citizen calls for snow removal services, and communicate with the drivers for information about each route. 

In order to keep snow removal costs manageable, such local governments as the City of Indianapolis, are using GIS tools along with new operation processes to make better decisions and plans for clearing the city streets of snow and ice. Since 1996, the City of Indianapolis has been using an innovative advanced GIS application called "Snow Fighter" that helps the Department of Public Works manage snow removal during winter emergencies. 

Snow Fighter is an application that overlays information about the snow removal activities on a map of the street network of Indianapolis. So, managers and supervisors from the Department of Public Works are able to visualize the amount of snow on the roads, the location of the snow trucks, and the amount of snow removed from the routes. Snow Fighter integrates a combination of software modules. The user interface and application management software is written in Visual Basic. This module also is used to interface with the Oracle database that allows the module to store and retrieve data that are collected from work orders and snow management forms during the project. The GIS functions for the application use ESRI's Map Objects. Internet map server (IMS) application updates information about the snow removal process on map coverages originally created using ArcInfo. The IMS also enables the application to be shared to all authorized users who are located in disparate office and garage locations throughout the city. The application uses real time data radioed in from the drivers as a key input into a networking algorithm to determine the snow routes that need to be plowed. The algorithm helps the managers to compute the optimal route for each truck plowing snow. The city is experimenting with using GPS and an automated data reporting system to track the real-time location and other attributes of each truck. 

This application has greatly improved snow removal services in Indianapolis. Snow Fighter allows managers in the department to develop a strategy for removing snow from the city streets and spreading traction enhancing materials (e.g. salt, sand, and chemicals), manage and track the actual snow removal, communicate with the public about snow removal progress, and submit information and forms to federal and state departments of emergency management for cost recovery. The Department of Public Works can remove snow more quickly, calculate the snow-removal costs quickly and transparently, and estimate the needs and costs of supplies and equipment more accurately.

III. Implications for Future Research

At the heart of these three advanced GIS applications are a number of management, technical and policy issues that may limit the broader adoption of similar E-government services in the near future. One of the key management challenges is training. Clearly training users to use the home-delivered meals systems and Snow Fighter must be addressed, particularly to reduce the number of data entry errors. The modeling modules create deeper need for training. These modeling tools are black boxes to most users and require specialized training on how to update, manage, and validate the models. Further research is needed to identify ways to design proper management and organizational controls to ensure that valid and reliable statistical, mathematical, and spatial models are used in advanced GIS applications. Other research initiatives should address: 

Additionally, there are a number of technical issues that are open for further research related to using GIS for E-government. Briefly, these advanced applications need to operate in a networked environment and enable access by users from all level of the organization. The applications also need to integrate data from disparate data sources and to distribute the information to not only desktop machines, but also other hand held devices. A brief list of important technical issue that may limit the adoption of these advanced applications include:

Database and related systems

Lastly, advanced GIS applications raise a number of public policy related issues. Using maps, especially delivered over the Internet opens up concern over privacy and security issues, and universal access.

REFERENCES

Gant, Jon, Wilpen Gorr, John Encandela, Dana Phillips, and Frank Wimberly, "Establishing Need and Designing Information Systems Supporting Decentralized Services for the Elderly: The Case of Allegheny County, Pennsylvania", Topics in Health Information Management, 1997, 19(1), pp. 1-10.