Uncertainty Analyses Of Regional Soil Erosion Model

Parameters of regional soil erosion models revelate the character of the studied region, is an important component of these models, and play an very important role in the model results. The forecast precision is the crucial target in appraisaling and weighting the model quality, and is also the determining factor whether a model is approved by the same profession personnel. Because the structure of models are unperfect, and model parameters, model initial condition and boundary condition data are insufficient, the precision of regional soil erosion models are low. The uncertainty research on regional soil erosion models became one of the discipline front fields in hydrology and soil science. Based on the regional soil erosion model which established by soil and water conservation expert-Yang Qinke, uncertainty analysize was conducted。Analyze the uncertainty and error source in model parameters, offer theory support for the model predictive error analysis; Analyze the sensitivity of the regional soil erosion model, determine the relations between definite model output variation and factor change. This can optimize model parameter, increase the model forecast precision and service its application and promotion.Based on analysizing the distributional basin hydrology model and comparaing domestic and foreign uncertainty research, results show that the indefinite error of the model came from 7 aspects: the error caused by removed factors , stochastic or system error caused by input data, error contained in historic record , error of unsuperior parameter value and parameter combination, error uncorrect algorithm produced , error imperfect model structure caused , error inappropriate time and spatial resolution, time step, interpolation and extension caused.Sensitive analysis of model input parameters (rain, anti, infi, DEM, LAI, time) was conducted using Perturbation Analytic Method。The other five parameters maintained the same, based on the starting value (each change 10%),each parameter was separately changed in the scope of positive and negative 40%。Then ran the model and analyzed the sensitivity of each parameter by observing object function's change. The conclusion initially obtained was:(1) The dominent raininess Rh parameter is rainfall amount, other parameters have no influence on raininess.(2) The main factors of maximum sediment deposition -Dp were: infi, anti, rain and time. Parameters sensitivity from strong to weak were: rain>anti>time>infi and each parameter was extremely insensitive to the Dp minimum value.(3) The sensitive parameters of runoff afflux Wflow were: rain and infi. Time step also had influence on the minimum value of Wflow. The influence of LAI on Wflow is very weak and extremely insensitive.(4) The influence factors of silt afflux Eflow were: rain, infi, anti and time. The sensitivity from strong to weak were: rain>anti>infi>time.(5) The sensitive parameter of runoff modulus Wm was: rain and infi. Time step was also sensitive to the minimum value of Wm .(6) The sensitive parameters of erosion modulus Em were: rain, infi, anti and time. The sensitivity from strong to weak were: rain>anti>infi>time.