Visualizing and quantifying microtopographic change of dryland landscapes from an unmanned aircraft system
Soil and site stability are key attributes of assessing the health of dryland landscapes because these lands are susceptible to high rates of wind- and water-caused erosion. Field techniques for measuring and monitoring soil erosion in drylands are often labor intensive and require many field visits. As a result, such field methods usually cover only small areas. There is a growing effort to monitor natural resources at broad scales that is driving the need for new tools to monitoring soil erosion and its effects on ecological functions and resource availability. One remote-sensing technique that could be used to measure soil movement from very-high resolution (VHR) aerial imagery is digital elevation model (DEM) differencing, where a DEM of the land surface is created using aerial photogrammetry methods at multiple points in time. By geographically co-registering DEMs created from VHR imagery obtained via unmanned aerial systems (UAS) and subtracting one elevation surface from the other, an estimate of soil elevation change can be defined. Such analysis allows quantification and visualization of soil erosion, deposition, and redistribution. We tested this technique to model the soil surface change of Chihuahuan Desert shrublands located near Las Cruces, NM, USA. We acquired 20 overlapping aerial images from a BAT 4 UAS for six 50 x 50m test plots in May 2014 and in February 2015. From the imagery, we created 5-cm spatial resolution DEMs of each of the plots and subtracted one image date from the other to create a difference image.
Accuracy of the soil surface DEMs was shown to be very good compared to ground-based elevation profiles generated from a laser range finder slid along an erosion bridge. 95% of aerial DEM height measurements were within 5 cm of the heights measured from the erosion bridge. The average difference between ground-based and aerial methods was less than 2 cm. Minor soil surface changes occurred between the image dates due to vehicular traffic and erosion processes. These changes were captured with strong agreement between the ground-based and aerial methods. Our results demonstrate that we can effectively observe the topographical change (cm scale) of rangeland soils from UAS platforms and gives distinct efficiency advantages for characterizing soil surface movement over broad areas. This information, available at scales that are commensurate with ecological processes operating in the study area, will allow for easier and more accurate quantification of resource distribution and loss.