The Technology Benchmark: A photogrammetry primer.
Based on these experiences, I was comfortable in my understanding of the technology. But I must admit I have not paid close attention to technological advancements in this field. In the past eight years or so, production capabilities have skyrocketed to the point that aerial mapping is a strong alternative to conventional surveying for many projects. Let’s take a quick tour through the science, practice and process of aerial surveying and photogrammetry. For definitions of the terms used in this article, see the glossary on page 66.
Modern MappingTo gather material for this article, I spent a day at McKenzieSnyder Inc. in Ashburn, Va., where I met with President Malcolm McKenzie and Vice President Jeff Snyder, whose combined experience in this field adds up to more than half a century. Founded in 2000, McKenzieSnyder performs mapping services for DOTs and other public agencies as well as private clients in land development.
As we discussed the types of projects appropriate for aerial mapping, I quickly learned that it is not just for large projects but is also effective on medium and even small projects. McKenzieSnyder routinely works on 3- to 5-acre projects and has been hired for commercial projects as small as a half-acre. The standard fee for a small project is less than $2,000, and the firm often delivers a completed project with both 2D and 3D data within a week. Therefore, McKenzie believes that the short turnaround time, the data accuracies captured and the low costs involved make photogrammetric services a strong alternative to conventional surveying.
Initial PlanningWhen clients approach a photogrammetric service provider, they typically outline their needs, provide a site boundary and state their desired contour interval. Although Snyder says his firm continually tries to educate clients that vertical accuracy is the key component to photogrammetric projects, the general customer is still primarily concerned with contour intervals.
This led me to ask Snyder and McKenzie what determines good vertical accuracy. According to McKenzie, there is no one explanation. Rather, it is a project-based analysis that includes cost considerations. Depending on the project’s size and budget, vertical accuracies can be obtained in several ways and cost efficiencies can be achieved by varying other parameters, such as ground control.
Snyder’s guidelines for success are to focus on project planning, which is the study of the client’s specifications and project analysis, and project design, which comprises the aircraft’s flying height and the amount of ground control to be attained. If you fly higher, you can obtain more coverage but accuracies are poorer. If you fly lower, your coverage is poorer but accuracies are higher. The trick is to fly the altitude that provides the best coverage with the best accuracies. For example, a flight height of 3,000 feet might produce the most cost-effective solution for a 1:500 scale project.
The Photogrammetric ProcessOnce the project is undertaken, the next step is to lay out the flight lines, or flight plan, based on the client-provided project boundary. McKenzieSnyder usually starts with a 7.5-minute USGS quad sheet and uses Flotron AG’s (www.flotron.ch) TopoFlight software program to help construct the flight line based on the desired scale. TopoFlight computes the optical characteristics of the lens and shows coverages and overlaps (see Figure 1 on page 64). The flight planning takes into account an overlap such that each photo overlaps the preceding and next photo by 60 percent with a side lap of 30 percent. Often, McKenzieSnyder underlays an existing orthophoto from USAPhotomaps in the workstation to offer a visual reference to the linework (see Figure 2 at left). Available at Download.com, USAPhotomaps is a free program that downloads USGS aerial photo and topo map data from Microsoft’s free TerraServer Web site.
McKenzieSnyder uses Intergraph (www.intergraph.com) ImageStations, which are digital photogrammetry workstations that allow the operator to set up the data within a much shorter time frame than in the past. (It used to be that each time an operator opened a project, he or she would have to spend about 15 minutes setting it up. Now the setup is stored with the project.) The flight plan is sent to the client with desired ground control locations. Because it is computed using a UTM coordinate system, the requested control points can be output using the WGS-84 datum. Visually Identifiable Objects (VIO) can also be used for control and located and requested, as well. The flight plan is output in a text format and provided to the flight crew to enter into their flight navigation software, which automatically directs the camera to fire.
McKenzieSnyder is often contracted for projects that can be sent to the aerial surveyor and, weather permitting, flown the same day. Over the next two to three days, the film is developed (if a conventional camera is used) and the photos are titled and scanned at 12 micron-pixels. (A micron is 1/1000 of a millimeter; there are 25,400 microns per inch.) These digital images are 150 MB each in compressed format. Once the digitals are received, they are input to the workstations.
Airborne GPS is another computation that can be used to benefit larger projects, according to McKenzie. The physical location of the aircraft and the camera and their relation to the principal point of the photo are established thereby making it possible to fly a 10,000-acre project with only five or six ground control points. This method also benefits projects involving mountains, wetlands, swamps and other hydrologically active areas.
From Data to DeliverableAfter capturing and processing the data, McKenzieSnyder performs data compilation with Bentley (www.bentley.com) MicroStation software. The software has a menu system that assists in assigning features, such as curbs, paved roads, driveways, sidewalks, transportation features, guard rails, shoulders, fences, walls, etc. McKenzieSnyder’s employees digitize the features and corresponding symbologies, a process that requires “visual spatialization,” or the ability to see the third dimension in the photos, to digitize accurately. “Human error can blow the project, and it is a very labor-intensive process. Every single point in the data set has been visited by the compiler,” McKenzie says.
“After the planimetric data is collected,” he continues, “the DTM collection process begins. Planimetrics consist of the two-dimensional features, while the DTM data contains all of the ground level features, such as deformations, breaklines, top and toes of slopes, etc.” The 3D data is then processed and run through a contour interpolation program. The technician then places spot shots and identifies high and low points. The process is repeated on the next stereo model. The size of the project and the amount of exposures dictates how many stereo models exist. Figure 6 below shows the result of the 3D DTM process: the plan with contours and DTM data.
When this work is completed, the project data is converted to a CAD file in the client’s desired format using Cardinal Systems’ (www.cardinalsystems.net) VR1 software, which assists in exporting stereo models into CAD format. “It is here that model joints (or edges) are evaluated," McKenzie explains. “Linework is joined together to create monolithic objects. For instance, a tree line that runs through multiple stereo models needs to be joined into a single element. Then cartographic editing occurs for text placement and grid placement, and layering and ground control are entered. This is all part of the quality control. If we see a tree line in one stereo model that ends at a joint and the next model doesn’t show the tree line, then we know someone missed digitizing the remaining part of the tree line.”
Upon completing this effort, the VR1 software exports to either a MicroStation or AutoCAD file. The deliverable typically consists of a planimetric file and the DTM file containing the raw 3D source data. However, many engineering firms do not realize that the source 3D data is already included in the deliverable. As a result, designers and technicians build their existing terrain models from the contour data provided instead of the more accurate source data on the deliverable CD. This oversight creates substandard terrain models.
From initial project planning to final deliverables, I hope our tour has provided a better understanding of how far advancements in the field of photogrammetry have taken the science, process and application of aerial surveying and mapping.
Photogrammetry GlossaryAbsolute orientation – the addition of ground control data to the data set.
Accuracies and Errors – usually within 1/10 of an inch for every 1,000 feet of flight elevation for a single point on a hard surface.
Aerial mapping – using photographs taken by flying above the site and reducing the information into 2D and 3D CAD data.
Aerial surveying – flying of the site and taking pictures.
Compiling – the act of digitizing the data.
Fiducials – accurate markings on the corners and midway points of the photographs.
Flight line – the centerline of the aircraft’s flight path.
Forward gain – the distance between the principal point in sequential photographs used to estimate how many exposures will be needed to capture a project.
Inner orientation – reading of the fiducials.
Overlap – the amount of information overlap with the preceding and next photo (typically 60 percent).
Panel points – the physical objects placed on the ground for control (typically a white cross shape).
Parallax – the apparent displacement of an observed object due to a change in the position of the observer.
Photo size – 9” x 9”.
Photogrammetry – the science of project analysis, flight line creation, photograph correlation, data capture and data quality control.
Relative orientation – the reading of Von Gruber points.
Side lap – the amount of overlap occurring between photographs to the left and right of each other (typically 30 percent).
Stereo model – photographs that are correlated and corrected.
Stereo pairs – two overlapped photographs.
VIO – Visually Identifiable Objects.
Von Gruber points – points established in the middle of the images.