Road Erosion And Its Influence On Gully Development In Loess Hilly Region

Soil erosion causes land degradation.As one of the indispensable components of the agricultural system,unpaved roads are significant sediment sources,but road erosion is often overlooked because of the relatively small areas that roads cover,and there are few studies to analyze the impact of road construction on hydrology and slope gully.In the Loess Plateau of China,road erosion occurs frequently,but the research on the mechanism of gully erosion on the road surface is weak.This paper aims to analyze the characteristics of road erosion under extreme rainfall conditions,determine the initiation threshold of road gully erosion,use landuse and rainfall data to predict road gully erosion on an event basis,and Clarify the influence of road on the development of gully.Provide scientific basis forroad erosion control and regional soil erosion prevention in loess hilly areas.On July 26,2017,an extreme rainstorm occurred in the Wuding River Basin of the Loess Plateau.We investigated road erosion in two small watersheds after a heavy rainstorm to measure rill and gully erosion on three types of road in the field and determine the factors affecting that erosion with remote sensing images and geographic information system（GIS）.After the rainy season in 2018,the roads in Shejiagou watershed were investigated again to analyze the erosion of unpaved roads after an extreme rainstorm of 212.2mm in 2017 and a regular rainstorm of 83.8mm in 2018.A digital surface model（DSM）derived from unmanned aerial vehicle（UAV）images was used toanalyze the road gradient（S）and upslope area（A）.The runoff(Q_RS)of each road segment（RS）was estimated by the runoff curve number method.Quantify the comprehensive threshold conditions of topography,rainfall and landuse in upslope drainage area required for the formation of gullys on the road surface,and propose an event-based model to predict the location and extent of road gully erosion.Quickbird image,UAV image and DSM were combined to analyze the influence of road on the characteristics,drainage area and runoff path of downslope gully.（1）The main form of road erosion in this study is gully erosion.On main unpaved roads,in which44.60%of the total road length in the studied watershed was destroyed,with an average soil loss of804.77 t/ha,equivalent to the average loss of a 5.28-cm-thick layer of soil.Gullies accounted for 92.3%of this erosion.On the three road types,the average length,width,depth and volume of gullies were larger than those of rills.The average rill erosion was 0.41,0.17,and 0.02 cm,compared with 4.88,3.15,and 1.05 cm of gully erosion on the main unpaved roads on secondary unpaved roads and trails,respectively.（2）Under heavy rainfall conditions,the main influencing factor of road erosion is drainage area.The single-factor linear regression model can explain 84.9%of the road segment erosion,with the equation y=243.469A^-15.432,R²=0.849.The nonlinear regression model using drainage area and road segment gradient as compound factors can better predict the occurrence of road erosion,and the equation is y=18.863A^1.837·S+2.563,R²=0.970.（3）The number of gullies and their dimensions were lower under aregular rainstorm than under an extreme rainstorm.The mean and total gully volume after the regular rainstorm of 83.8mm were,respectively,26.28%and 8.08%of those after the extreme 212.2mm rainstorm.The majority of gully formation on the surveyed roads could be explained by the threshold condition（S-0.071）A^2.211≥0.0005,whereas the threshold condition（S-0.056）Q_RS^3.363≥72.444 accounted for why many more road segments experienced gully erosion underthe extreme rainstorm event than under the regular rainstormevent.By incorporating the S–Q_RS threshold condition,the present study proposed two models for predicting gully erosion in a road segment（y）,namely a power function using the event runoff(Q_RS)as the predictor(y=0.003Q_RS^1.917+15.852,R²=0.960)and a non-linear regression model using event runoff(Q_RS)and road gradient（S）as predictors(y=0.003S^-0.140Q_RS^1.889+15.273,R²=0.962).（4）The road on the loess slope significantly affects the flow path.In the three cases of no road impact from 2007 to 2019,road impact from 2007 to 2019,and no road impact in 2007 but impact in2019,the changes of average gullyheads width and drainage area are 0.44m,0.97m,1.76m and164.61m²,435.28m² and 1229.88m²,respectively.The average and maximum increase widths of gullyhead with road influence were 4 times and 6.5 times of that without road influence in two years,respectively.21.74%of gullyheads that had no road impact in the past two years did not move forward,while there was no road in 2007 but road in 2019,the shortest distance to go forward was 1.60m.With the influence of road,the runoff on the road surface converges to the roadside gap and acts on the gullyhead of the slope with a single flow path.When there is no influence of road,the runoff acts on the gullyheads with multiple scattered paths.