Redefining Spatial Boundaries
May 1, 2006
One of the most common problems land surveyors face is the re-establishment of site control that has been destroyed or removed. The ability to spatially coordinate and manage surveys within a GIS environment that simultaneously unifies, integrates and perpetuates surveys with survey-grade accuracy is an attractive idea to many. Recovering what can be decades-old site control to perform resurveys can be challenging. And assembling and analyzing volumes of data from prior surveys to efficiently perform new surveys compounds the work effort.
For National Survey & Engineering, a division of R.A. Smith & Associates of Brookfield, Wis., the likelihood that completed surveys will adjoin current projects is high; this is because the firm has completed a large number of surveys. But the readying of current projects means time-consuming assemblage of old notes and paper plats of surveys that have a bearing on the lines of the new surveys. It means extensive analysis of that data. And when it comes time to relate this information to what is on the ground, the sparse control leads to an even bigger challenge: most of the old surveys that affect the new boundary solution exist in spatial isolation. That is, they are not easily connected to each other through common lines, coordinates or basis for bearings. After all, what surveyor has intentionally traversed between the numerous surveys he has done with the expectation that the accurate spatial alignment between them might be needed to perform a future survey?
The network of more than 1,000 Continuously Operating Reference Stations (CORS) in combination with National Survey & Engineering's ability to write value-added code for modern GIS software has radically impacted the firm's ability to meet these challenges.
A Solution to ConsiderNational Survey & Engineering brought its survey and GIS groups together to develop a high-accuracy survey-grade GIS solution called Boundless. The patent-pending software extracts point and line information from standard electronic drawings and point databases, and transforms it into a unified coordinate system that does not scale or alter geometry established and measured on the ground by legal description.
Boundless comprises value-added code written for ESRI (Redlands, Calif.) ArcGIS, and operates on survey drawings and databases tied to the National Spatial Reference System (NSRS) using high-accuracy static GPS adjusted by the method of least squares to the CORS network.
Boundless does not hinder, disrupt or alter in any way customary office procedures employed within survey departments to produce plats of survey and to manage point databases. Survey project managers and CAD drafters at National Survey & Engineering do not need to do anything to operate or maintain Boundless.
How It WorksAll surveys using Boundless are done using GPS to tie the points and lines of the survey to the International Terrestrial Reference Frame (ITRF), a common earth-centered, earth-fixed (ECEF) coordinate system. Even if total stations are used on the project, this link to the ITRF is necessary. The surveys may be produced for clients and other users as they always have been, tied to local control. But the underlying data is on a unified coordinate system. For this to be practical, the ITRF coordinates need to be projected onto a plane coordinate system, usually Transverse Mercator or Lambert. However, most state plane coordinate systems and the Universal Transverse Mercator (UTM) introduce scale differences. This requires surveyors to convert distances found by inversing between plane coordinate systems to ground distances. Sometimes this conversion is overlooked, leading to errors.
Taking a page from the Wisconsin County Coordinate System developed by the Wisconsin Department of Transportation, Boundless uses projections over a fairly small area where the elevation of the projection surface lies very close to the actual ground elevation of the county. Thus, the scale differences are greatly reduced, making the discrepancy between projected and ground distances well within the tolerances for most kinds of local surveying.
The location of a GPS base station used for surveying a project can now be established by logging only a few hours of data at the base station and then post-processing with multiple CORS as far as 100 miles away. The CORS data, along with precise orbits, can be easily obtained over the Internet. The processed vectors to three or more CORS can be adjusted by the method of least squares holding the published CORS coordinates fixed. The result is extraordinary horizontal positional accuracy.
Although it is possible to very accurately position all surveys in relation to each other and perpetuate those positions relative to the CORS, most surveys are still produced and drafted on individual coordinate systems that have no intrinsic spatial relationship with or to any other survey or coordinate system. But they do represent geometry as measured on the ground. The problem then becomes how to simultaneously transform multiple drawing files and databases into a common system that retains ground distances as grid distances, and where the relationship between points and lines on the different surveys now all relate to each other. A further challenge is how to accomplish all this using standard field techniques, drafting methods and equipment while not disrupting office procedures or requiring special training.
Thus part of the solution involves using high-quality GPS control for each survey. Since GPS has its own universal coordinate system worldwide, the "native" coordinates derived for all points in two or more separate, previously untied surveys will now be on a common basis. However, even surveyors have difficulty easily relating to the GPS-based coordinates. Projecting the GPS coordinates in such a way as to not introduce scaling differences (i.e., the difference between grid and ground distances are minimal) is the second part of the solution. If two separate surveys are to be spatially unified using GPS, and at the same time not introduce scaling, then a projection must be used that minimizes the difference between projected grid distances and ground distances; these differences should be virtually insignificant and well within the magnitude of errors associated with the best practices of surveying.
Unlike state plane coordinate systems, county coordinate systems largely minimize grid/ground differences by confining the projections to smaller areas, that is, counties instead of whole regions of a state. For much of Wisconsin, the Wisconsin County Co-ordinate System reduces the difference between projected grid and ground distances to a level that falls comfortably within expected survey measurement errors over the distances that would be encountered. The region in which a property boundary solution could be expected to influence other boundary solutions is reasonably within a square mile. For the eastern counties of Wisconsin, the difference between county grid distances and ground distances over a mile is on the order of one- to two-hundredths of a foot. This is well within even ALTA/ACSM standards for survey measurement accuracy, which allow for 0.07 feet relative positional accuracy at the 95 percent confidence level (or 20 mm + 50 ppm). Where no official county systems currently exist such as in Illinois, National Survey & Engineering has created county coordinate systems for its own use.
If a minimum of two points from a survey are tied to the CORS network and held in a county coordinate system database, and if these same two points exist in the drawing database used to produce a survey, then a relationship exists between the coordinate system in which the survey is produced and the county system. Transformation of a survey drawing file and associated database into the county system now becomes a simple translation and rotation defined by the point pair common to both systems. In this way, otherwise unconnected surveys can be spatially united without scaling or distorting the geometry of a completed survey that has been measured and platted based upon ground measurements.
GIS-aided SurveysGIS software is the ideal engine for spatially transforming (translating and rotating to bring the points into the correct mathematical relationship with each other) multiple surveys. The proprietary value-added code developed by National Survey & Engineering for ArcGIS automatically identifies, extracts and correlates common points in different coordinate systems by accessing Microsoft Access databases used by Autodesk (San Rafael, Calif.) Land Desktop Development (LDD) and Trimble (Sunnyvale, Calif.) Geomatics Office (TGO).
For each survey drawing and point database chosen for transformation, the code searches for and finds two tagged points in a TGO project database in the Wisconsin County Coordinate System. It then searches for and finds the same point numbers in the survey database used to draft the survey. The corresponding coordinate pairings representing the same physical points but in two separate coordinate systems define the unique transformation applied on-the-fly to each point and line in each survey.
A typical survey to be included in Boundless might go from start to finish as follows (we use Trimble GPS and robotic total stations; reference will be made to this equipment and software):
1. A one-man survey "crew" begins by setting up a GPS base station to do RTK while also logging satellite data at the base for eventual post-processing to multiple CORS. The surveyor begins the survey using no projection or datum. The surveyor locates section corners and establishes site control using RTK. During this process the surveyor also performs a one-scale site calibration, thereby creating a local Transverse Mercator projection for his site that applies a scale factor so his grid coordinates are at ground. After this is complete, he occupies the RTK points with a robotic total station and proceeds with data collection.
2. In the office, the data is imported into TGO along with all static data from the rover and base station. High-accuracy static vectors between the base and rover may be post-processed to displace and supersede the less accurate RTK vectors that were computed in real-time during the survey. After the combined RTK, static, total station network data has been analyzed, checked and perhaps adjusted by least squares, the point information is imported into LDD for drafting.
3. The TGO project is then copied and renamed to a standard format. RINEX files for at least three CORS are imported, post-processed and adjusted by least squares onto the appropriate county coordinate system. The point descriptions of three points in the TGO project are altered by appending "WGS84" to one, "GPS1" to another, and "GPS2" to the remaining point. The latitude and longitude of the WGS84 point determine the graphic location of the dot representing the survey within ArcMap. The northings and eastings of GPS1 and GPS2 define the translation and rotation to be applied to the survey drawing files and point database to convert them to county coordinates.
4. Next, a proprietary routine called Survey GIS Project Manager (SGPM) is used to find the point with "WGS84" appended to it. It extracts the latitude, longitude and county for this point and puts it in a Microsoft SQL Server database, including the project number.
Now an ESRI ArcMap file (.mxd) containing the value-added code is opened. ArcMap is then used to update the Shape file containing survey project information. To do this, Boundless searches the SQL database for the point with "WGS84" appended to it and places a dot on the map representing its location.
5. To transform a survey to county coordinates and put it on a unified basis with other surveys, Boundless finds "GPS1" and "GPS2" in the county TGO project database. It then compares the coordinates of the same points in the survey database (they will be on two different coordinate systems, and thus the values will be different). The translation and rotation parameters are then calculated and applied to the specified lines and points. ArcMap, of course, can be used to then display the now-related surveys. Multiple surveys can be selected for simultaneous transformation in the foregoing manner with a process time measured in seconds.
6. After drawings and points from any number of surveys have been selected and unified with high survey-grade accuracy, the unified point and line information can be selectively exported as Shape and point files for import into new project databases for use in office and field systems. Because there will likely be point numbering conflicts between the survey point databases that are unified, the system consecutively renumbers the exported points. To maintain traceability after export, it appends the source project number and original point number to the description field of each exported point.
Because the transformation of a survey to a county coordinate system is defined by an association that depends on points that exist within a survey drawing database, translating and rotating that database never disrupts subsequent transformations. In this regard, the system is intrinsically self-updating. This was a central software design requirement for National Survey & Engineering because, in the process of producing a survey, it is not uncommon for a drawing and associated database to undergo several translations and rotations in an attempt to arrive at the final boundary solution.
Satisfying another design requirement, the solution allows older surveys with point databases to be grandfathered into the system by simply tying two prior control points to the NSRS via static GPS. It averages about 15 minutes to download data from four CORS and post-process to a base station at a survey site, including the time to perform a least squares adjustment. Once this process is complete, the survey becomes a part of Boundless. Any further field work, whether conventional or GPS, that finds its way into drawings and associated databases will be available for transformation the next time the program is run. GPS data can be collected for inclusion of a survey in Boundless either before standard field measurements, during or after the other field measurements have been collected. The number of points that must be located in order to include a survey in Boundless is minimal, allowing crews to leave dual-frequency receivers running unattended while they perform other tasks. The net result is a robust system that requires very little extra effort or maintenance.
A Case StudyA recent ALTA/ACSM land title survey illustrates the application of Boundless. Miller Compressing in Milwaukee, Wis., owns a large (one mile east-west) heavy industrial site where cars are crushed and recycled for scrap. Here, control points have a short life span and many property lines are determined by building corners and rail lines.
Two separate prior surveys, which had located most of the needed section corners, were in Boundless. Because of this spatial merging of the section corners, some pre-boundary surveying analysis could be performed in the office. This analysis in turn assisted the crews in establishing the lines of the new survey. The crews only had to occupy one control point from one of the older surveys with an RTK base station to begin working within a pre-established coordinate system.
To complete this survey, we had to connect to property boundary control that included control established or found in adjoining or nearby surveys we have done over the years. Had all those old surveys been in Boundless, the effort (and cost) of establishing Miller's boundaries would have been substantially reduced. We would have been able to unify the older adjoining surveys and retrace them on the ground undeterred by the loss of physical evidence long ago replaced by new sidewalks, curbs and development.
Our current survey of Miller Compressing is tied to the NSRS and perpetuated within Boundless and thus will have a direct, high-value positive impact on future surveys in the area.