新疆玛纳斯河流域洪水的实时预报研究

Manas River Flood Prediction Study

在改进的自回归模型AR(p)基础上,运用最小信息准则确定模型的阶数p,建立玛纳斯河流域的洪水预报模型。根据新建立的模型,在充分考虑了融雪径流型河流自身特点的基础上,对其进行参数的动态修正,选择出了与实际情况相吻合的最佳参数。该模型率定期和检验期确定性系数均大于0.9,径流总量相对误差均在1％以下,最小为0.34%,洪峰合格率为95.0％,超过国家水情预报规范中的甲级标准(合格率≥85％)。

Considereding characteristic of snow storage and snow melting on Manas river, use autoregression model AR(p) and minimum information(AIC) criterion,caculate the best steps(p) of model, carry on the dynamic revision of the parameter, to reach the best result of the parameters.Determinacy coefficient of this model is greater than 0.9,surface flow total amount relative error is below 1% ,minimum 0.34% ,qualification rate of flood crest 95.0% ,the first rate standard exceeds the national water level while predicting normally (qualification rate 85%).