The Use of Stereoscopic Satellite Images to Map Rills and Ephemeral Gullies

Accurate mapping and measurement of erosion channels is necessary to accurately estimate the impact of channeled erosion in an area. Field surveys can provide optimal quantitative results, but they are only applicable to small areas. Recently, photogrammetric techniques have been applied to small format aerial photographs that were taken by UAVs. Few studies have applied photogrammetry for mapping and measuring single permanent gullies using very high resolution stereoscopic satellite images. We explore the use of such images to map rills and ephemeral gullies and to measure the length, width and depth of individual erosion channels to estimate the eroded volumes. The proposed methodology was applied to the Collazzone area of Central Italy. All of the channel characteristics were determined using GeoEye-1 panchromatic stereoscopic satellite images of the 48-km study area and a 3D floating cursor. We identified, mapped, and measured the lengths of 555 channel segments. The top width and depth could be measured in only a subset of the channel segments (the SMC subset). The SMC data were used to determine the coefficients of the power law relationship between the rill/gully volume and length (V = aL) and the uncertainties due to the channel depth measurements and the cross-sectional shape. The field data of the rill and gully volumes were within the estimated uncertainty. We defined a decision rule to distinguish rills from gullies on the basis of the segment length and applied the corresponding power law relationship that was derived from the SMC subset to estimate the eroded volume of the entire dataset. The erosion values that were calculated at different scales (0.680 Mg∙ha at the catchment scale, 28.4 Mg∙ha on the parcels affected by erosion) are consistent with values found in the literature. Our results indicate that erosion OPEN ACCESS Remote Sens. 2015, 7 14152 at the catchment scale can be considered moderate, whereas the erosion at the field scale exceeds the tolerance limit, which is consistent with data that have been summarized and/or discussed by several authors.