The principal aim of the paper is to determine which index of random roughness is most appropriate for the description of the microtopographic variations within a rainfall series at the interrill scale. The study was conducted with rainfall simulation on interrill-sized plots.The roughness indices were computed from the point cloud or from the Digital Elevation Model (DEM) of the soil surface, obtained from photogrammetric surveys. The indices evaluated can be grouped into five categories:‘contact method – pin’ and ‘contact method – chain’ where the indices are obtained from the DEM simulating manual techniques (pin meter and chain respectively); ‘fitting plane/surface’, where the indices are the standard deviation of the height differences from a best-fitting plane/surface of the DEM or of the cloud; ‘moving windowDEM’, where the index is the average standard deviation of the residual topography; and ‘kernel size cloud’, anew index given by the standard deviation of the distances between each point of the cloud and a best fitting plane computed from its nearest neighbors (found in a radius called “kernel size”).The performance of the indices was assessed based on their relationship with the measures of cumulative rainfall and kinetic energy, depression storage, cumulative runoff and soil loss. The best index to represent random roughness variations is that of the ‘moving window DEM’ category with a resolution of 0.0025 m2, RI25,for which the relationships with the cumulative rainfall, the cumulative rainfall kinetic energy and the depression storage, have coefficients of determination of 0.86, 0.85 and 0.80 respectively. For the same relationships, a good but slightly lower performance is provided by the indices of the categories ‘contact method –chain’ and ‘kernel size cloud’. These three categories gave equivalent results regarding the correlation with cumulative runoff and soil loss (mean Pearson correlation coefficient −0.95, CV = 0.025). Instead, the indicesof the categories “fitting plane/surface” and “contact method – pin” resulted poorly correlated with all the measured variables
A comparative evaluation of random roughness indices by rainfall simulator and photogrammetry
A. Vinci
;F. Todisco;L. Vergni;
2020
Abstract
The principal aim of the paper is to determine which index of random roughness is most appropriate for the description of the microtopographic variations within a rainfall series at the interrill scale. The study was conducted with rainfall simulation on interrill-sized plots.The roughness indices were computed from the point cloud or from the Digital Elevation Model (DEM) of the soil surface, obtained from photogrammetric surveys. The indices evaluated can be grouped into five categories:‘contact method – pin’ and ‘contact method – chain’ where the indices are obtained from the DEM simulating manual techniques (pin meter and chain respectively); ‘fitting plane/surface’, where the indices are the standard deviation of the height differences from a best-fitting plane/surface of the DEM or of the cloud; ‘moving windowDEM’, where the index is the average standard deviation of the residual topography; and ‘kernel size cloud’, anew index given by the standard deviation of the distances between each point of the cloud and a best fitting plane computed from its nearest neighbors (found in a radius called “kernel size”).The performance of the indices was assessed based on their relationship with the measures of cumulative rainfall and kinetic energy, depression storage, cumulative runoff and soil loss. The best index to represent random roughness variations is that of the ‘moving window DEM’ category with a resolution of 0.0025 m2, RI25,for which the relationships with the cumulative rainfall, the cumulative rainfall kinetic energy and the depression storage, have coefficients of determination of 0.86, 0.85 and 0.80 respectively. For the same relationships, a good but slightly lower performance is provided by the indices of the categories ‘contact method –chain’ and ‘kernel size cloud’. These three categories gave equivalent results regarding the correlation with cumulative runoff and soil loss (mean Pearson correlation coefficient −0.95, CV = 0.025). Instead, the indicesof the categories “fitting plane/surface” and “contact method – pin” resulted poorly correlated with all the measured variablesI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
