|
|
|
| Home | Projects | Members | Publications | Committee | Data | RSPA | Meetings | Contact |
| Background | |
| Overview |
|
| Considerations | |
| Case Study | |
| Conclusions | Return to Environmental Assessment Home |
|
Characterization of surface topography is necessary for transportation corridor planning. Accurate terrain information is vital in determining alignment and grade. Used in conjunction with volumetric analysis tools, this same information allows estimation of cut and fill operations. Analyzed in light of local geology and hydrology, it can be used to assess landslide potential. Biogeographic analysis can also indicate potential ecological impacts as a result of corridor construction and utilization.
Common practice in corridor planning involves the use of surveying and/or photogrammetry to generate topographic information. Data collections by surveying requires a considerable investment of time and labor and may not be practical for large projects. Crews also face potential hazard exposure in the field. Conversely, the creation of a photographic database for photogrammetry is hindered by considerations of weather, sun angle, and vegetation condition. Consequently, photo acquisition is typically limited to early spring and late fall. Additionally, while generally not as labor intensive as surveying, photogrammetric methods involve significant time and labor costs. As an adjunct to these methods, the creation of a digital elevation model (DEM) from remotely sensed data can provide savings of both time and labor while accelerating the early planning stages of a project dealing with corridor route alternatives and their potential benefits, impacts, and drawbacks.
|
|
One of the most efficient means of gathering remotely sensed elevation data is by interferometric synthetic aperture radar (IFSAR). IFSAR is a single pass method in which an aircraft utilizes two antennae separated from each other by a known distance in the cross-track direction. As in a single antenna radar system, energy is emitted toward a target and reflections of that energy are received by the antenna. IFSAR represents the slight difference in return times between the two receivers as a phase shift which can be processed ("unwrapped") to calculate a distance from the platform to the target. By factoring in precise aircraft positional data, elevation information is derived and a digital representation of the target surface produced.
This basic first surface product, called the digital surface model (DSM), represents the relatively unprocessed dataset which includes the return signal from various hard objects such as buildings and trees that project above the earth's surface. Beyond serving as the basis for a finished DEM, the DSM is particularly useful as an underlying topographic structure for digital image draping. By utilizing high resolution digital aerial photography or images from satellites such as QuickBird, photorealistic three dimensional models of study areas can be produced quickly and economically.
In most instances, however, the transportation planner will need a bare surface product, which is the DEM. Using automated spatial processing tools, surface objects are removed from the elevation dataset. Several checks are incorporated into the processing sequence to avoid removing real, high relief surface features which may mimic the appearance of the objects being filtered out. Final quality checks and local editing complete the transformation from DSM to "bald earth" DEM.
|
|
From a transportation planning point of view, DEMs produced from IFSAR data have both desirable and undesirable characteristics. Because of the wavelengths at which the instrument operates, data collection may proceed with little consideration of weather. Clouds, fog, smoke, and ambient temperature have no effect on the radar signal. This simplifies and eases the task of mission planning, especially in contrast to the high regard which must be accorded weather conditions when planning aerial photography acquisition.
Also attributable to the operational wavelength, however, is a sensitivity to tree canopy density. Once the radar signal enters the canopy, it can be deflected and/or reflected. Some of the signal will strike the top of the canopy at near normal angles and be reflected back to the receiver. Another portion will pass through the canopy, strike the surface, and be reflected back unimpeded. Some portion of the signal will be deflected away and lost. The remainder of the signal will enter the canopy and be reflected one or more times within the trees before ultimately returning to the platform. So long as the signal from the ground and the canopy top are distinct, they can be differenced to derive an accurately modeled ground surface. However, with a strong enough return from the signal that "rattled" within the canopy, detecting the true ground return can be difficult. IFSAR data can be collected very quickly from an aerial platform. Because the instrument can capture signal phase change, a single pass is sufficient to gather the necessary information used to derive elevation. As opposed to aerial photography where a rapidly moving platform can have adverse effects on image quality, the IFSAR platform's velocity causes no appreciable degradation. In fact, IFSAR instruments are commonly mounted on jet aircraft in order to acquire large datasets as quickly as possible. Cost of acquisition is low, as well. For areas of 10,000 km², a DSM can be had for approximately $2.00 per km².
DEMs produced from IFSAR excel at providing extremely large topographic datasets at very low cost. In their current state, however, these DEMs provide neither the accuracy nor the precision of either photogrammetry or surveying, especially in large scale study areas. Depending on the particular dataset purchased, typical RMSE values range from 0.7 m to 1.0 m, with mean errors of 0.5 m to 0.7 m. 95% of the values will fall within a range of ±1.5-2.0 m of the true elevation.
|
|
In this project, we examined the potential utility of DEMs derived from interferometric synthetic aperture radar when used as a planning tool for transportation corridors in a typical mixed forest that might be encountered in the southeastern United States. Our goal was to test the limits of accuracy and precision of traditional IFSAR DEMs in order to determine the smallest scale at which they might be useful for planning and assessment.
The metropolitan Atlanta, Georgia, area was chosen to be the test site. Intermap Technologies was contracted to acquire the IFSAR data in February, 2001. The entire dataset was collected in a single mission flown at approximately 6100 m (20,000 ft) above mean ground level. The resulting DEM is a grid of 2.5 m pixels derived from 5 m posting data. Nominal accuracy was specified to be 2.0 m vertical and 2.5 m horizontal. Initial quality assessment of the resultant dataset indicated problems with vertical accuracy in forested areas. The most severe errors were in broadleaf deciduous forest land. Errors were still apparent but less severe in mixed broadleaf deciduous and coniferous evergreen forests. Some mild to moderate error was noted in pure stands of coniferous evergreens. These errors were ultimately ascribed to the radar signal bouncing among various leaf layers between the forest canopy and the ground. These multiple bounces appeared to the instrument as increased distance from the platform to the ground. After consultation with Intermap, an effort was made to systematically correct for the vertical errors. Initially, a data set was sought that would provide elevation ground control points for the most affected areas. The intent was to incorporate these control points into Intermap's automated surface modeling routine in order to provide a processing bias toward known good elevations. Unfortunately, no such data set was readily available. At our request, Intermap then performed a manual correction process over a subset of the data in order to assess the viability of such a routine in filtering anomalous elevation values. Ultimately, we concluded that hand corrections would not be economically viable in a large scale corridor planning scenario in a forested environment.
|
|
Digital elevation model production from IFSAR data can be a valuable tool for transportation planners. In this case study, a vary large elevation database was compiled quickly and economically. Automated processing provided a quick and generally accurate model of the topography of the study area, a model suitable for use in consideration of corridor alignments. Specific problems of elevation accuracy in this study could have been considerably alleviated by acquiring the IFSAR data during cooler weather when the deciduous trees would have dropped their leaves and the evergreen needle coverage would have been less dense. In areas lacking thick forest cover, such considerations would be unnecessary. Indeed, many IFSAR-derived DEMs exist for areas without forests and they routinely meet or exceed desired accuracy levels.
Minor accuracy issues aside, however, IFSAR data can be collected and a DEM compiled quickly and relatively cheaply early in a project. This DEM can be utilized to determine a set of corridor alignment alternatives for consideration. Once final corridor selection is made, only that corridor's area would require coverage by surveying or photogrammetric techniques. By using this two step approach, large tracts may be excluded from the time and labor costs associated with producing the high accuracy and precision terrain data necessary for route final project engineering.
|