Municipal planners in Atlanta, GA, are integrating Geographic Information Systems (GIS) with customized problem-solving computer models to project the impact of future land use and urban development on the city’s stormwater and sanitary sewer systems. Armed with this valuable information, they are creating a watershed management program that will help Atlanta decision-makers develop the best plan for the future.
The graphic and mapping capabilities of GIS are strong tools for conveying information to the public and government officials. GIS has helped advance research and analysis of hydrologic processes and how spatial patterns affect water resources. Introduction of this technology into city and county government provides opportunities for increasing government efficiency. The majority (80%) of data managed by cities and counties is location-related.
The computer-based GIS stores digital data representing features on the Earth as points, lines, or polygons and maintains an attribute database for each feature. For example, land uses, soils, rainfall data, watershed boundaries, land cover, and underground infrastructure all can be mapped, attributed, and stored in GIS.
At a minimum, the various features can be overlaid for analysis. GIS also provides spatial analysis tools (based on the spatial relationship principles of connectivity, contiguity, proximity, and the overlay technique) for manipulating geographic data. These functions give the GIS user a way to work with spatial data to answer questions about geographic relationships.
At a higher level, GIS can be integrated with discipline-specific models. Data relevant to hydrologic/hydraulic modeling can be extracted from the GIS databases and modeled without the user leaving the GIS application.
GIS tools are becoming increasingly valuable for water resource management because they can perform multiple “what-if” modeling of the watershed system after the data are collected, stored, and managed. For example, alternative conceptual designs for improving stormwater runoff can be created with minimum data management and time consumption. This is in addition to its extraordinary ability to coordinate massive amounts of information.
Atlanta’s Watershed Project
The City of Atlanta contracted HDR/WLJorden Inc. of Atlanta in November 1999 to develop a GIS-based decision-support system for watershed management. The objective of the system is to provide a comprehensive approach to development and infrastructure planning for projects within the city.
The GIS-based decision-support system framework initially supported the East Area conceptual design of the combined sewer separation. The approach considers future land use and urban development within the watershed basins, separates the storm and sanitary sewer conveyance systems, and provides the additional stormwater and sanitary sewer conveyance capacity to improve water quality and the natural environment.
The high-end use of GIS is to provide support for making spatial decisions. In current GIS environments, basic GIS operations (data storage, management, manipulation, and analysis) provide the tools for generating data inputs for spatial decision analysis. To provide a richer decision-support environment, it is necessary to integrate analytical models with GIS software and an easy-to-use graphical user interface.
The decision-making process has been described as a dynamic feedback loop consisting of three elements: opportunity identification, alternative(s) generation and design, and choice.
The “opportunity identification component” determined the need to separate the combined sewers in Atlanta. During this phase of the project, data are acquired and stored in the GIS database.
Next, the design component of the decision-making process involves developing a set of possible alternatives to fulfill the identified opportunity. With respect to the conceptual sewer separation design, a traditional engineering method was used for compiling relevant data in Microsoft Excel spreadsheets and AutoCAD drawings (a situation where the tabular data are not connected to the drawings).
In this workflow process, the design engineer begins with a sewer system map, projects sewer and stormwater flows for a given catchment and service area by hand, and designs the structure of the separated system. Proposed sewer and stormwater flows are modeled in the Storm Water Management Model (SWMM). The input data for the SWMM are compiled from spreadsheets into an input format for manholes and pipes relevant for the model.
The engineering design group had two external constraints: (1) a six-month time frame to go from data compilation to a conceptual design with estimated costs and (2) multiple data sources. The data validity had not yet been proven, and the sewer mapping was outdated. It was obvious that the data would need to be constantly validated and updated as conflicts were resolved and new information was received. Creating conceptual designs under these conditions and in such a short amount of time with the standard workflow process seemed impossible. Consequently, a new set of tools called HDRLink was developed.
New Tools for Modeling
HDRLink was created to be a software link between ArcView GIS and Hydra modeling software. Environmental Systems Research Institute of Redlands, CA, produces ArcInfo and ArcView software. Hydra, developed by Pizer Inc. in Seattle, WA, is commercially available hydrologic and hydraulic software that is similar to SWMM.
Doing the conceptual design within GIS and a linked hydrology/hydraulic model is a fundamentally different workflow process than that outlined above. The model requires user inputs about the collection system and flows that enter the system and then calculates flows and hydraulic conditions throughout the network. The data are compiled and linked to the mapping in one step. This fundamental feature of GIS relieves the pressure of data management in a constantly changing data environment. When data change or sewer lines and manholes need to be edited as a result of new information, that change can be made once and is automatically reflected in both the database and the mapping.
The model then can be used to evaluate possible improvements to the system. This approach gives maximum flexibility to modeling hypothetical scenarios to produce several alternative stormwater and sewer system designs. With a trained staff, most of these projects can be done in-house and on an as-needed basis. Alternative conceptual designs can be created with maximum data management. (The data management headache of the older workflow process cannot be overemphasized in terms of greater time consumption and possible corruption of data.)
The “what-if?” question embodies the choice component of the decision-making process. HDRLink enables GIS data to be exported to a hydrology/hydraulic model that emulates stormwater and sewer infrastructure conditions. It then allows the user to export the model’s results back to the ArcView GIS environment. The strength of ArcView GIS is its adaptability for application development that enables geographic data to be combined with any modeling software.
Project Overview
The following sections will describe the phases of framework development leading up to the integration of the GIS stormwater modeling–from gathering different data sources to setting up the data for the hydraulic model.
GIS data sources are seldom intentionally designed for hydraulic modeling. Typically facility maintenance programs, land-use planning, and other department programs are responsible for the development and maintenance of the data sources used for their purposes. The objective of the GIS/model interface is to translate the GIS information to the format required by the hydraulic model and include model output processing such as costing information.
The focus in the development of a GIS interface for a hydraulic model should be to utilize existing data sources that are typically maintained or will be maintained by other departments. During the data compilation and modeling process, data from each department is rigorously analyzed and checked for integrity. Communication between departments and the consultant is crucial in ensuring that inconsistencies and anomalies are corrected.
Sewer System Models
To run a computer model of a combined sewer system, two things are necessary: a collection system and the wastewater/stormwater flows that run through it. Data collection provides information about the collection system. Wastewater/stormwater flow assignment is accomplished based on land-use patterns. With information about the land use in a given area, the land-use/flow extraction process can estimate the amount of wastewater/stormwater that enters the system at a given point. Point flows (industrial facilities) are identified and can be added to the model on an individual basis. The model also can account for wet-weather flows (infiltration and inflow from high groundwater levels and rainfall).
The input information necessary to develop a combined sewer collection system model can be divided into three broad categories:
- Collection System Information: Information such as pipe sizes and lengths, manhole elevations and locations, and pump sizes.
- Land-Use Information: Information about current and future land-use trends in the area of study.
- Flow Assignment Data: Includes information to calculate and relate land-use-based flows, point flows, and wet-weather flows to the collection system.
HDRLink flow assignment options (as configured for Hydra) include assignment of flows with diurnal flow patterns. Supported flow sources include land-use-based flows, point-source flows, wet-weather flows, and storm-related infiltration. The Flow Assignment Wizard allows the same base data, such as land use, to be used to generate model scenarios for different growth conditions and design events. It also includes a flow-factor calibration option, translates the land-use data directly to flows, and at the same time assigns a service area-specific diurnal curve to the inflow. This assigned curve is selected based on the land use that contributes to the flow.
After a model is run and calibrated, the calibrated Hydra output file is loaded into the HDRLink-Hydra interface, creating an output file that contains hydrographs for every pipe in the system and a maximum hydraulic grade line (HGL) elevation for each pipe. The HGL can be recalculated for each 15-minute time step in the hydrograph, giving a more detailed assessment of the system’s hydraulic condition.
Data Collection
Combined Sewer Collection System
The major data sources for the East Area combined sewer overflow project and the GIS decision-support system were the City of Atlanta planning and public works departments. The public works department provided paper documents containing information on the existing combined sewer system. These documents included archive plans; field books and files dating back to the 1920s; and city sewer maps, rainage maps, and design plans prepared by various firms contracted by the city over the years.
Some information existed in electronic form (Excel spreadsheets and sewer map files) and represented partial sewer or combined sewer separation pipeline and manhole data. The city’s sewer maps were accurate circa 1996 (before construction for the Olympic games).
The majority of the data had to be converted from hard-copy maps and tables to digital data before any data manipulation or analysis could occur. In addition to compiling existing wastewater pipe and facility infrastructure data, existing and future land-use data had to be converted into GIS themes. Those hard-copy maps were provided by the planning department. This was time-consuming but necessary for using HDRLink. In evolutionary terms, the data-compilation part of the project went from a one-celled organism to a human being in six months.
Land Use
Existing land-use GIS themes were based on a digital building footprint mapping layer obtained from the public works department. Time and budget constraints prohibited developing the building attribute database from information from the tax assessor’s office. A proprietary source for parcel-level tax assessor information (which included fields for land use and zoning) was used. However, this information was found to be out of date in many instances. In lieu of accurate tax data, a field survey was conducted to verify such building attributes as the number of floors.
The future land-use maps were digitized into the GIS from hard-copy neighborhood planning unit (NPU) maps obtained from the city planning department. The NPU future land-use maps define the maximum development densities for each parcel in the city. The future land-use densities allowed the engineering team to determine future capacity requirements for the separated sewer system and for stormwater runoff. The percentage of impervious surface was based on land use. The acres of impervious surface were calculated in the GIS using the future land-use density data in the storm catchment areas.
Flow Assignment
Sewer and storm catchment areas were defined prior to generating numbers for existing sanitary sewage flow rates and stormwater volumes. The extent of the catchment areas was about 1 ac. each. Once these areas are defined within the GIS, HDRLink allows the designer to assign the flows generated within that catchment area to any particular injection point (manhole). The flows are calculated based on the catchment area and the land use within that area.
As stated earlier, the land-use layer was derived from a base building footprint coverage provided by the public works department and a building database compiled from a commercial tax data source and a field inventory. Developing the land use from current building information provides parcel-level information regarding the type of land use generating sewer/stormwater flows. The database contained specific information about large commercial buildings and multifamily and single-family residences in the study area. The information gathered about the commercial buildings included the number of floors in each building. In cases of multifamily residences, the data gathered included the number of units in each building. Finally, a count of single-family residences was computed from the database.
Using 1998 digital orthophotography provided by the public works department, the building footprint shapefile was updated by adding major buildings that were surveyed. Also, buildings that have been demolished since development of the shapefile and that were field-verified as no longer existing were deleted.
Estimating Sewage Flow Rates
The combined sources of data were used to generate square footage of commercial buildings, unit numbers of multifamily housing, and total numbers of single-family dwelling units. These were generated for each catchment service area and are the basis for estimating existing sewer flows and stormwater volumes.
The existing square footage of commercial area for each catchment service area was calculated by multiplying the square footage of the building footprint by the total number of floors in the building. The number of units of both multifamily housing and single-family residences was calculated by selecting the multifamily or single-family buildings within the service area and summing the units.
The databases are structured so that as more building data are collected, the database is updated and flows can be regenerated using updated information.
The future estimated flow rates were calculated based on the future land-use map. This was accomplished by using the delineated service areas to select a section of parcels. The land use, zoning, vacancy, and acreage are kept in an inventory database for each of the parcels. Using this information, a total for each of the categories was determined and an ArcView Script was run that summarizes the existing data into one table. This table contains the available information about the summarized parcels in that service area. For example, Table 1 shows one of the service areas in a sub-basin and the groupings that resulted from the land-use combinations.
Table 1. Service Areas
| Number of Parcels | Acres | Land Use | Zoning | Status |
| 23 | 2.99 | Single-Family Residential | R4B | Developed |
| 1 | 1.15 | Low-Density Residential | RG3 | Developed |
| 1 | 0.09 | Single-Family Residential | C3C | Developed |
| 7 | 0.89 | Single-Family Residential | C3C | Vacant Lot, Residential |
| 56 | 5.44 | Single-Family Residential | R4B | Vacant Lot, Residential |
| Total | 10.56 |
For the Atlanta project, future development focused on the buildout potential of vacant parcels. This approach assumes that major downtown buildings will remain during the planning period. The parcel-level approach to building the land-use layer provides the basis for analyzing individual (large) developments at the parcel level.
The development of this wide range of spatial data combined with the HDRLink tools has added a substantial increase in the utility of the data and the GIS. Data updates can be made relatively quickly and easily. The results from the changes or additional information also can be updated much easier than with traditional methods that use separate software for graphics, data storage, and modeling.
Conclusion
GIS-HDRLink is a good data management system because it requires a one-to-one correspondence between the graphic feature, such as a sewer line or storm sewer segment, and a database record. This implies a complete inventory of assets. The system is also very flexible, as it allows easy editing of geographic databases so that as data are updated, they are immediately updated throughout the system and can be remodeled seamlessly.
The system also provides a high level of decision support, above mere thematic mapping, by linking an engineering design model to the geographic data that provide inputs to the design model. It is an excellent environment for modeling alternatives and for visual presentation of geographic data and design alternatives that laypeople can absorb.
There are upfront investments in GIS software that must be taken into consideration with this system. HDRLink costs $5,000, ArcView costs $1,000, and Hydra costs $3,000. There are also expenses to be expected for data conversion, training, and maintenance. As part of the Atlanta project, HDRLink will be adapted to link with SWMM.
