Assessing soil erosion of agriculture field during winter and summer seasons using 3D scanning and close-range photogrammetry technology

Abstract

Soil surface morphology is strongly affected by land management practices and soil biophysical processes such as soil erosion. Traditional methods on measuring soil surface morphology changes have been costly, time-consuming and causing extra soil disturbance. Alternatively, remote sensing technologies involving 3D scanning and photogrammetry are becoming available and capable of measuring surface morphology correctly and quickly. With these techniques, studies have focused on small-scale rill erosion during short snow-free periods. In this study, the performances and accuracies of 3D scanner and photogrammetry methods on detecting morphological features, i.e., total station scanner (TSS) and close-range photogrammetry (CRP), were evaluated and optimized. Three parameters (width, depth and area) of each morphological feature of two surfaces, a ridged and channeled plywood board (2.4 m by 3.6 m) and a ridged and channeled bare-earth plot (6 m by 20 m), were derived using both remote sensing methods and actual measurement. For the plywood board assessment, the optimal TSS and CRP root mean square errors achieved were: width = 1.3 cm vs. 0.8 cm; depth = 1.0 cm vs. 0.4 cm; areas = 7.8 cm[squared] vs. 4.2 cm[squared]. Whereas for the bare-earth plot, the optimal TSS and CRP root mean square errors were: width = 3.4 cm vs. 1.1 cm; depth = 2.2 cm vs. 1.6 cm. Hence, the performances of CRP on detecting morphological changes are better than the TSS results. However, the TSS method is more practical than CRP by allowing larger scanning distance and fewer operators. A field experiment was conducted at the Agriculture and Agri-Food Canada Fredericton Research and Development Centre, New Brunswick, on 6 m by 80 m large plots to determine surface morphological changes under three tillage treatments during two winter and summer seasons. The treatments involved (i) potato cropping with up-down-slope tillage, (ii) potato cropping with contour tillage, and (iii) fallowing with up-down-slope tillage (control). The periods lasted (i) from snowfall to after snowmelt, and (ii) from seeding to about one-third potato canopy coverage. Surface morphologies (elevation, slope, curvature) were scanned and evaluated at the beginning and end of each period. Also, soil moisture at 15 and 30 cm depth, and soil temperature at 15 cm depth were monitored within the winter period before and after freezing. For the winter period, there was an about 2.1 cm to 2.9 cm drop in elevation for the plots as a whole. Translating these changes in soil losses, assuming that these changes were due to soil erosion, and that the soil bulk density equals to 1.1 g cm[to the power of -3], they can procure a net soil loss of 231 Mg ha[to the power of -1] to 319 Mg ha[to the power of -1]. However, this elevation changes is unlikely being caused by soil erosion alone, but attribute to soil-structure and compactions caused by freeze-thaw action, snowpack compaction that could change soil aggregation and pore volumes. For the summer period, the plots revealed an elevation drop in the distance ranges of approximately 0 m – 20 m and 40 m – 50 m from the top of the plots. On the other hand, elevation rise was observed at approximately 20 – 40 m and 50 – 70 m. This drop and rise would translate into overall soil loss of 55 to 110 Mg ha[to the power of -1] for the cropped plots, and a net soil gain of 22 to 88 Mg ha[to the power of -1] for the fallow plots. In some of the winter and summer plots, water-induced erosion rill was detected using both TSS and CRP methods. For the same rill derived from the summer period, CRP yields higher and more accurate value than TSS by about 0.9 Mg ha[to the power of -1]. Overall, these results indicate that both technologies are able to capture surface elevation changes and detect erosion rills. However, neither method could be used to determine sheet erosion correctly. Hence, further methodology refinements are needed to ensure that the TSS and CRP results are not affected by the freeze-thaw action or the presence of ground-covering vegetation due to, e.g. cropping or fallowing.