以绿地为下垫面,分析了覆盖度等因素对径流系数的影响。结果表明:覆盖度低于80%时,草地的下渗能力较小,高于80%时显著增强,径流系数与覆盖度成非线性关系。为了准确预测绿地的产流量,根据多种影响因素与径流系数构成的多维非线性关系,提出采用-ε支持向量回归机建立绿地径流系数的预测模型,与LM-BP神经网络预测模型进行了比较。结果证明:-ε支持向量回归机建立的径流系数预测模型具有泛化能力强和预测精度高的特点,为城区绿地产流量的预测提供了新的计算方法。
Taking the greenbelt as underlying surface, the influences of cover degree on runoff coeffieient were analyzed. The results showed that the infiltration capacity was less as the cover degree under 80% and it was increasing evidently as; the cover degree over 80%. The relationship of runoff eoefficient and cover degree was nonlinear. To predict the runoff amount accurately, the ε-support vector regression model based on the multidimensional nonlinear relationship of various influence factors and runoff coeffieient was presented and compared it with ANN model, The results indicated that the ε-support vector regression model has the properties of strong generalization capability and high forecast preeision. It provided a new method for foreeasting the runoffproducing amount of city.