Print PageA study was conducted at the Jornada Long Term Ecological Research (LTER), New Mexico with the purpose of determining the impact of wind erosion on paired 25-m X 50-m plots where 100% of the grass cover had been removed compared to the control plot. The objectives of our study were to compare manual measures of plant number, height, density, cover, and biomass of a 3-D point cloud of each plot to a semi-automated method that measures the rendered point cloud of each plot, and 2) characterize the differences in 3-D vegetation structure and soil surface microtopography after 4 years of wind erosion events. Point clouds were generated with a 4.5-mm spot resolution 524-nm terrestrial scanning laser (TSL). We hypothesized that wind erosion has had a homogenizing effect on the 100% removal plot. We also expect that vegetation height diversity and other parameters such as density and number will be less on the 100% removed plot.
Results/Conclusions
We collected 4 TSL scans per plot including GPS readings on 3 survey targets. Scans overlapped and were all 30 degree vertical and 180 degree horizontal with a 40-m radius. Point Clouds were registered and exported as x,y,z text files for each plot to various software packages including Surfer, Quick Terrain Modeler (QTM), R, Idrisi, and TreeVaW, for point cloud and grid volume analyses. QTM was used to remove vegetation and generate ground surface digital elevation models, and to manually measure the vegetation parameters of the virtual plots. TreeVaW was used to semi-automatically measure number of stems, plant height, and canopy cover. We found that surface roughness on the 100% removed plot < the control plot, convexity on the 100% removed plot < the control plot, and flatness on the 100% removed plot > the control plot, because micro- and macro- hummocks have formed from sediment entrapment on the control plot. We found significant differences between site vegetation with greater plant numbers, cover, density, and biomass on the control plot. We had slight differences in vegetation parameter estimates being dependant on the pixel resolution chosen for gridding with less similarity in estimates and increasing overestimates the coarser the pixels become. We found that the base resolution is comparable to field selection of nested plot sizes in shrublands ~ 49 cm2, probably the system length scale.
