Loading...
HomeMy WebLinkAboutGIS Pilot Study, Final Report and Maps GALLATIN COUNTY/CITY OF BOZEMAN GIS PILOT STUDY FINAL REPORT AND MAPS Prepared by: Jackie Magnant and John P. Wilson Geographic Information&Analysis Center Montana State University Bozeman, MT 59717-0348 27 September 1995 BACKGROUND In the preliminary stages of Geographic Information Systems (GIS) implementation in a local government setting,the most critical steps usually involve: (1) the acquisition of accurate base layer data,and (2) staff training in software and data familiarity. Both steps will have major impacts on the success and integrity of the GIS. This pilot study was designed to research the most effective method to accomplish these tasks while crafting a blueprint for successful implementation of GIS in Gallatin County and the City of Bozeman. Sections 13 through 32 in T2S R6E were selected in consultation with city and county staff and used for the investigation of data sources and data conversion options as part of this pilot study (Figure 1). METHODOLOGY Four basic approaches were used to prepare digital files (i.e., ARC/INFO coverages)for the study area as follows: 1.Preparation of a GPS Road Survey: The centerlines of major and secondary roads were surveyed with Global Positioning Systems (GPS)receivers. The equipment that was used consisted of a Trimble receiver attached to a rooftop antennae, a laptop computer with the Geolink mapping software, and a base station. The Geolink software drew the roads data in vector format(i.e.,as a series of lines defined by pairs of x,y,z coordinates) on the laptop screen as the data were being collected by the receiver and the laptop operator entered attribute data which was saved in a database specially designed for this project (see Appendix A). The base station is located on the top of Leon Johnson Hall at Montana State University-Bozeman and its location has been surveyed by the Civil 2 Engineering Department to submeter accuracy. 2. Conversion of Gallatin County Road Rook Man Sheets: The recently updated Gallatin County road book portrays roads, streams, subdivisions, Certificates of Survey (COSs),land parcels, street addresses and other types of annotation on approximately 400 different map sheets. Map sheets 1302 through 1305 (which cover the study area) were obtained as AutoCad DXF files from Morrison and Maierle,Inc. (Gallatin County road book contractor), imported into ARC/INFO, and converted into a COLO coverage (i.e., spatial database). 3. Preparation of GIS Data Layers from Certificates of Survey(C.OS.s:The bearings and distances recorded on subdivision descriptions and COSs were entered into ARC/INFO and used to construct a text file(containing a digital copy of the source data) and an ARC/INFO COGO coverage. The COLO module incorporates geometric tools for calculating parcel boundaries and other features directly from legal descriptions and survey data. 4. Conversion of City of Bozeman AutoCad File: These files, which were prepared in AutoCad by the City Engineering Department from the bearings and distances reported in subdivision descriptions and COSs, were imported into ARC/INFO and converted into a COGO coverage as well. The preparation of these different data layers served multiple purposes. The road survey provided a 1-5 meter accurate road base layer and four additional GIS layers consisting of milepost markers, hydrants, bridges, and signs. A series of control points were also created from the road layer and these data provided a ground control layer for the placement of subdivisions, COSs, and future layers (see Figure 2). Finally, the GPS data provided the modem coordinates for the coverages produced from all four data sources. The road book map sheets were copied from AutoCad to ARC/INFO because they contain large volumes of pertinent spatial and attribute information in a digital format and they might have eliminated the need for a road survey and/or some other (large) data collection effort. The COLO coverage prepared from scratch in ARC/INFO(method 3) was needed to determine: (1) the accuracy of the features recorded on the road book maps relative to the locations reported in the original surveys, (2)the relative proportion of vector and raster formatted data, and (3) the utility of importing the road book data into the ARC/INFO GIS and using some or all of these data as a base data layer. The final method was needed to determine whether or not the drafting software currently used by the City Engineering Department (AutoCad Release 12) produced the same results as the COGO module in ARC/INFO. The similarities and differences between the three types of COGO data and the original surveys were calculated mathematically in ARC/INFO and a series of visual comparisons were used to compare the GPS road survey results with the other(COLO) data products. 3 Two city/county staff worked extensively on the data collection and data conversion tasks to: (1) facilitate data transfer, (2)learn more about software options and data entry, database design, and data editing and manipulation issues, and (3) help with the evaluation of data sources and data conversion methods. RESULTS AND RECOMMENDATIONS There are at least two ways to present the results of this pilot study. One approach would focus on the search for the best approach for building a base layer for a county- and/or city-wide GIS and the other would simply explore the value of automating the different data sources that were examined to construct one or more GIS data layers. These approaches overlap because the GPS road survey provides numerous data layers as well as the only way to build a cost-effective GIS base layer with known locational accuracy(quality). The detailed results presented below describe the advantages and disadvantages of using the different data sources and/or conversion options for generating a variety of GIS data layers. The GPS roads survey (method 1) provided a roads coverage that can be used for geocoding, 911, and other road-based GIS applications. In addition, the: (1) large number and variety of GIS applications based on street addresses, (2) the accuracy of the surveyed positions (the other data conversion options provide locational data of variable and unknown accuracy whereas differential correction of the locations collected with the roving GPS receiver can be expected to deliver 1-5 meter horizontal accuracy),and(3)the use of modem coordinate systems for GPS surveys (the GPS receivers capture locational information in geographic (i.e., latitude/longitude/altitude) as opposed to some local coordinate system) mean that a road coverage would provide an excellent base map (data layer) on which to build a city/county-wide GIS. The results from repetitive surveys of several road segments demonstrated that we could resurvey the road system to within± 1-2 meters (Table 1). Finally, the design of the database was modified several times during the course of the pilot study to accelerate data collection and cleanup (back at GIAC lab), and these savings are incorporated in the proposal now under review (see new contract proposal). The results of converting the Gallatin County road book map sheets to a series of GIS data layers (method 2)would not provide a very satisfactory GIS base map because of the following problems: 1. The GPS road survey results were needed to transform the road book information into a modern coordinate system. This is an important requirement because most GIS tools (buffering, overlay tools, etc.) and therefore the benefits of implementing GIS require locational attributes stored in modern coordinate systems (see Wilson (1995) for further details and examples). 2. The boundaries of subdivisions and COSs on the road book map sheets did not always match those produced with COLO directly from the survey data (method 3) (see insert A in Figure 3 for examples). 4 3. Map sheet 1305 did not align properly with map sheet 1302 or the roads from the GPS survey (see insert B on Figure 3). In addition, individual subdivision boundaries did not match the COLO data when map sheet 1305 was shifted to overlay directly on top of the COGO subdivision data because of a pervasive problem with map sheet 1305 (the boundary of map sheet 1305 was larger than 1302) and subdivisions and COSs would need to be extracted individually and/or resized to match those on the adjacent map sheets. 4. Map Sheets 1303 and 1304 were obtained from Morrison and Maierle, Inc. in a raster format with no attributes and these data would need to be converted to a vector format before implementing this particular data conversion option. The conversion of the road book map sheets(method 2) does represent one of two ways of building a parcel GIS data layer. Digital road book files would need to be imported into ARC/INFO and converted to a COGO coverage. Missing features (parcels)could be entered in COLO from original survey data. Features in raster format would need to be converted to a vector format or reentered from the original surveys. A visual examination of these files could be used in most instances to determine which method is quicker. The time and effort required would also vary with level (intensity) of land development. We spent 7 hours building a parcel layer for map sheet 1302 and 10 hours building a parcel layer for map sheet 1305 for example. The road book map sheets also provided street addresses and several other types of annotation(see Figure 2 for an example). These data could be quickly imported into one or more GIS data layers and matched with the street addresses recorded in the digital Gallatin County Tax Assessor Office files in many instances to provide an immediate link to the CAMAS database. The preparation of a GIS data layer directly from survey data in COGO (method 3) provided a second option for generating a parcel data layer. This approach would provide the a more accurate parcel layer but will require more time compared to the road book option. However, this method will not maintain the original survey data precisely because the results achieved by entering the bearings and distances directly from the original surveys varied with the accuracy of the bearings and distances recorded by the surveyor. Errors may be present in the original survey data because: (1) many surveys are very old (which means the survey instruments that were used are not as precise as modem state-of-the-art equipment), (2)reference monuments have been resurveyed (shifted) numerous times, (3)different surveyors may have used different methods, and(4) some measurements may have been calculated and/or recorded improperly. These errors resulted in no closures of the traverse (which is necessary to delineate subdivision, COS, and individual lot boundaries) or boundary overshoots (see insert D in Figure 3 for examples of these problems). The no closures problems can be rectified by adding or deleting an arc and tagging the arc to indicate variations from the original survey measurements. Overshoots can be resolved by adjusting the traverse using one of the three methods of adjustment available in ARC/INFO (they all distribute errors over the traverse resulting in changes to direction and distance). It was also necessary to specify a correction angle for each township when entering the survey data (bearings and distances) with this method. Most of these problems could be mitigated by adopting appropriate database standards and data conversion protocols prior to building 5 the database. Another advantage of this method is that it did allow the legal descriptions for the subdivisions, COSs, and deeds to be saved in a digital text file. However,this data conversion option is probably more expensive than the road book option given that it took approximately 1-2 hours to enter a small subdivision in COGO, 3-4 hours to enter a medium-sized subdivision, and 5-6 hours to enter the bearings and directions for a large subdivision. A more complete analysis of the potential benefits and costs of using either the road book or original survey information as a data source for one or more GIS data layers was not possible because only four out of a total of approximately 400 road book map sheets were available for the pilot project. There is also the problem of identifying required applications once the city- and county-wide GIS is constructed since the preferred or required level of locational accuracy for a parcel data layer can be expected to vary with different types of applications. The AutoCad procedure currently used by the City Engineering Department to build a parcel data layer is conceptually very similar to that available in ARC/INFO (i.e., method 3 above). This approach, which is attractive because it capitalizes on existing equipment and staff skills in the engineering department, would provide greater long-term benefits if it was modified in two important ways: (1) finding and entering at least four registration points (tics)for each map sheet (AutoCad drawing file) (two of the four files used in the pilot study lacked distinct features that could be used to attach control points and transform the data to a real world coordinate system for example), and(2)the bearings and distances differed slightly from the actual survey data and/or the ARC/INFO COGO results because of registration errors noted above and the use of single precision instead of double precision for data capture and storage in AutoCad. Both of these problems can be avoided in the future by choosing double precision for data representation and entering the control points from the completed GPS road survey in the appropriate AutoCad files. The direct involvement of the appropriate city and county staff in this pilot project was beneficial from at least three points of view: (1) fleshing out the data needs and database design options, (2) accelerating data collection and editing (because of staff familiarity with data), and (3)providing GPS and GIS training for city and county staff. These outcomes have been incorporated in the new proposal to help keep costs down and to accelerate the development of GIS applications that assist city and county government. SUMMARY The preparation of an accurate digital road coverage for the City of Bozeman and Gallatin County represents a logical first step in building a GIS database to assist city and county government. The results of the pilot project indicate that the GPS hardware, Geolink software, and ARC/INFO GIS software can be combined to generate a digital road centerline coverage (1-5 meter horizontal accuracy) and several additional data layers (milepost markers, hydrants, bridges, road signs, culverts, etc.). Ground control points may also be created from the road layer and used to locate subdivisions, COSs, and other features in real world coordinates. 6 In addition, these survey control points (and double precision) should be adopted by the City Engineering Department to accelerate the development of a parcel data layer for the city. A closer examination of the digital copies of road book map sheets is required to know more about how the county might build a parcel data layer. Choosing between the road book and COLO options (methods 2 and 3 in pilot study)has important implications in terms of both database development costs and final data quality. The final decision will be affected by the intended uses of this data layer as well. The continued involvement and training of city and county staff in future projects will accelerate both the development of GIS databases by GIAC staff and students and implementation of GIS in city and county government. REFERENCES CITED Wilson, J.P. 1995. Reinventing local government with GIS. Public Works, 127(5): 38-39, 93. APPENDIX A ROAD NAME STARTING ADDRESS ENDING ADDRESS NUMBER OF LANES IN ROAD ROAD TYPE SPEED LIMIT DIRECTION OF SPEED LIMIT CONDITION OF ROAD MATERIAL TYPE CULVERT INTERSECTION NAME INTERSECTION TYPE TRAFFIC SIGNS BRIDGES OFF RAMPS MILEPOST MARKER NUMBER MILEPOST MARKER ELEVATION HYDRANT HYDRANT ELEVATION TABLE 1 GPS SURVEY RESULTS ON SELECTED ROADS LOCATION MINIMUM MAXIMUM MEAN SOUTH THIRD .323 1.58 .960 I-90 7TH TO PARK .326 2.136 1.003 COUNTY LINE SOURDOUGH .600 1.824 1.023 GARDNER PARK .948 2.112 1.15 I WILLSON .688 2.2 1.48 This table defines the differences in meters on individual road segments that were surveyed more than once.