Simulating Rill Erosion Process On Sloping Surface Based On The Cellular Automata

Studying on the spatial-temporal dynamic simulation of slope rill erosion processes is the current soil erosion discipline frontier science problem, and is also one of the hot topics in China and abroad. This study, aiming at slope rill erosion processes, using the cellular automata theory and method, combined with numerical simulation, GIS, and 3D laser scanning technique, coupling rainfall, infiltration, runoff, erosion, sediment transport and deposition in a unified CA framework, established CA numerical model for simulating slope rill erosion processes. Using the artificial simulated rainfall experiment to explore runoff sediment yielding characteristics under the process of rill erosion occurrence, development and evolution on the same slope surface by multiple rainfalls experiment; Understanding of the process of rill erosion profundly, providing theoretical basis and data support for establishing rill erosion model. The model simulation results were vertified under the same conditions,and evaluating the accuracy and reliability of the established model, using artificial rainfall simulation experimental data. The conclusions were as follows:(1)This paper constructed rill erosion processes CA model by extending CA standard structure, Defining cell space, model hypothesizes conditions and boundary conditions, determining the appropriate cell size, neighbor configuration and model assumptions, coupling rainfall, infiltration, runoff, erosion, sediment transport and deposition in a unified CA framework, determining the rules of infiltration, overland flow forming, soil particles move and deposition on each operating link. This model can estimate the development trend of rill erosion under different rainfall situations, output rill net and slope rill erosion DEM data, and make the rill erosion process visualization avaliable.(2)In the process of rill erosion on loess 15 o slope, feature of runoff and sediment yield under 7 rainfalls shows: Overviewing the seven rainfalls, the process of each rainfall runoff and sediment production presents the trend that increases first and then keeps stable. Only the first rainfall are slightly different: At the beginning of the first rainfall, moisture of prophase soil is low, Water infiltration is fast, which makes amounts of loose soil particles on the slope surface under the action of raindrops striking. Then surface runoff begins After 1.44min. With the extension of rainfall duration, surface soil water content gradually reached saturation, runoff rate increases; Thereby sediment transporting caused sand transport rate to increase quickly to 107.52 g/(m2?min); With the surface soil grain decrease, sand transport rate dropped to 73.92 g/(m2?min); Along with the increase of rainfalls, rill formed gradually, runoff rate tended to be stable. However, due to constantly fluviraption and sapping, sand transport rate increased steadily and with some fluctuation.(3)Dynamicly monitoring rill morphological development process of slope rill erosion with the 3D laser scanner. It indicates that the rill erosion is a complicated process by analysising the result, the process of rill morphological development formed as: small hydraulic drop--incised gully head --intermittent rill--continuous rill—rill network. During this process the rill gully head kept cutting ward and deepening continuously, and the collapse of gully cliff led to the widening of the rill, the gully head headward erosion and the rill lengthening. Meanwhile the rill was presented as the form of continuous bifurcation and combination, and eventually the rill network was formed gradually. Because the loess 20 o slope is longer, the rill net formed 40min later, and the rill net would be mature 60min later. The rill of loess 15°slope developed slowly compared with loess 20°slope, so we can monitor the whole process of the dich erosion completely.(4)Verifing simulated total rill erosion by CA model using runoff and sediment data obtained from artificial rainfall. Independent Sample T-tests are conducted for simulation value and the experiment value, which indicates that the differences between the simulating value and experimental value are not significant for loessial soil 25 o slope and loessial soil 15 o slope. There is a little difference between the simulating value and experimental value, only several simulating total rill erosion value is unreliable. The relative error range is 3.3% ~ 26.2%. All the 7 rainfalls of loessial soil 15 o slope relative errors were less than 15% as well as loessial soil 25 o slope except the last rainfall's 26.2%. Generally speaking, the simulating values are closed to the experimental value, therefore we can assume that this model can simulate the total erosion of per rainfall.(5) Verifing simulated rill morphological development process by CA model using the 3D laser scanner for scanning results. By contrasting experiment to the simulation result of erosion parameters, we may draw such a conclusion: CA model can completely simulate the slope rill morphological development process; But results of parameters simulating the rill erosion loessial soil 15 o slope wasn't ideal, the relative error is large; However, while simulating lou soil 20 o slope, 39% erosion parameters'relative errors are less than 10%; 50% of the relative errors are between 10% ~ 20%; 50% of the relative errors are more than20%. Simulation value and the experimental value are similar, And only a little value's relative error relatively bigger, so we can firmly believe that the result of simulated loess soil 20 o slope by CA model is more reliable.