Using landscape metrics to analyze micro-scale soil erosion processes

Authors: MARCO DA SILVA, ALEXANDRE - Sao Paulo State University (UNESP); Huang, Chi Hua; Francesconi, Wendy; SAINTIL, THALIKA - Florida A & M University; VILLEGAS, JAZMIN - Northeastern Illinois University

Submitted to: Ecological Indicators
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/7/2015
Publication Date: 9/1/2015
Publication URL: http://handle.nal.usda.gov/10113/61153
Citation: Marco Da Silva, A., Huang, C., Francesconi, W., Saintil, T., Villegas, J. 2015. Using landscape metrics to analyze micro-scale soil erosion processes. Ecological Indicators. 56(2015):184-193.

Interpretive Summary: Water movement on the earth surface shapes the morphology of the landscape. Photographs taken on the landscape allow scientists to use different image attributes to quantify the specific matrics of the landscape. For sample, plants on the surface can be considered as patches and their spatial distribution reported as patch size, shape and density. Likewise, topographic feautres can aso be quantified through a similar matrix. In this research, an analogy is proposed to see whether digital images taken from a soil surface at the micro-scale before and after rain events can be quantified using the same landscape attributes designed for topographic analysis at the macro or landscape scale. In a soil box that was subjected to a sequence of rainfall events, we found that the landscape matrix were capable to capture the significant changes on surface micropotography from the first erosive rain when erosional features were developed. Subsequent rainfall events caused minimal changes in the surface microtopography, hence little changes in the landscape matrix used in this study. This research shows that techniques used in landscape analayis can be down-scaled to quantify changes on the surface micromorphology resulted from erosion processes occuring at one erosive rain events.

Technical Abstract: Methods of recording soil erosion using photographs are common but they are not commonly considered in scientific studies. Digital images may hold an expressive amount of information that can be extracted quickly in different manners. One of these manners might be through the quantification of several metrics that were initially created for landscape analysis under an ecological context. The study is a first attempt to apply the method of landscape metrics to quantify spatial configuration of surface microtopography and erosion-related features, in order to generate a possible complementary tool for environmental management. In a 3.7 m wide and 9.7 m long soil box used during a rainfall simulation study, digital images were systematically acquired in four instances: (a) when the soil was dry; (b) after a short duration rain for initial wetting; (c) after the first erosive rain; and (d) after the 2nd erosive rain. Thirteen locations were established in the box and digital images were taken at these locations with the camera positioned at the same orthogonal distance from the soil surface under the same ambient light intensity. Digital photographs were converted into bimodal images and analyzed for seven landscape related metrics: percentage of land, number of patches, patch density, largest patch index, edge density, shape index, and fractal dimension. The digital images, as sources of information, were considered satisfactory because they can be acquired very quickly. The metrics were sensitive to changes in soil surface micro-morphology especially after the 1st erosive rain event, indicating significant erosional feature development between the initial wetting and first erosive rainfall. This analysis is a suitable method for spatial patterns of soil microtopography evolution from rainfall events that bear similarity to landscape scale pattern evolution from eco-hydrological processes. Although much more study is needed for calibrating the landscape metrics at the micro-scale, this study is a step forward in demonstrating the feasibility of the method.