Measuring Erosion Rates at Plot Scales Using Photogrammetry

Abstract

Listen

Translate article

Measuring erosion in natural environments can be complex, time consuming, and labor intensive, but several computer software programs have recently become commercially available that may allow accurate calculation of eroded volumes from digital photographs. These programs may greatly simplify the collection and analysis of erosion data under certain conditions. The objective of this project was to determine the accuracy of erosion estimates made with one photogrammetry software package (Adam Technology, Perth, West Australia) relative to estimates derived using more traditional methods. We compared the photogrammetric and traditional methods in two laboratory experiments and one field setting. The photogrammetric method consisted of obtaining a series of stereo digital photographs using a high-resolution camera and creating digital terrain models using the software package. The volume change between the before and after digital terrain models was also calculated using the software. The first test was a simple trench (approximately 10 cm wide by 4 cm deep by 24 cm long) created in a laboratory setting. The digital terrain model for this event was derived from two overlapping photos. In this case the volume calculated using the photogrammetric method was larger than the volume calculated using a rill meter (487 cm3). The second experiment was a runoff event in a 0.4 m wide by 11.4 m long laboratory flume. The digital terrain models for this event were developed from two sets of 29 overlapping photos. The volume calculated from the photogrammetric method was compared to the volume change between the before and after surface elevations measured using a scanning laser surface profiler (0.278 m3). The third test was a simulated runoff experiment conducted on a 9 m long plot in a burned area where two sets of 21 photos were used to create the photogrammetric models of the rills created during the experiment. The volume calculated from the digital terrain models was compared to an estimate calculated from mass of the eroded sediment collected in runoff samples and the soil bulk density (0.0412 m3). Obtaining each photo series required some pre-planning and the images were more easily processed in the software when the surfaces were lit with mid-day sunlight and few shadows. The photogrammetric method may allow researchers to rapidly and cost-effectively obtain relatively precise digital elevation models in field or laboratory settings, and to capture small changes in surface elevation.