水文模拟不确定性研究是当今水文科学研究中的热点问题。本文以流域水文模型SWAT为例，以黑河流域作为研究区，基于贝叶斯理论和方法，探讨SWAT模型参数不确定性对流量模拟和预测结果的影响。结果表明，在黑河流域，降水的变化幅度越大，模型参数不确定性对流量模拟和预测的影响越小；在降水变化相同条件下，温度增加会使模型参数不确定性对流量预测值的影响增大；在模型参数不确定性影响下，预测流量的不确定性区间夏季(尤其是6、7月份)最大；随着降水的减少预测流量值趋于减少，但预测流量值的不确定性区间变化不大；随着气温的升高，春季流量预测值有所增加，且春季流量预测值的不确定性区间也趋于增大；降水增加条件下，随着气温的降低，预测流量过程线越来越尖耸；降水减少条件下，随着气温的降低，预测流量过程线形状变化不大。 Uncertainty issue in hydrological modeling is a hot topic in recent hydrological research. Taking Heihe river basin as the study area, we mainly focused on the uncertainty in runoff prediction resulted from SWAT model parameter uncertainty. Bayesian method was employed for nine sensitive parameter estimations. We used 95% CI of runoff prediction to illustrate the uncertainty in runoff prediction caused by model parameter uncertainties. Results showed that: the larger the range of changing in precipitation, the narrower the 95% CIs of runoff prediction, the less effects of model parameter uncertainties to runoff prediction. With the increasing of temperature, the 95% CIs of runoff prediction were stretched if the precipitation kept stable, which means that the increases in temperature would lead to larger effects of model parameter uncertainty to runoff prediction. The 95% CIs of summer runoff prediction were the largest, followed by those of autumn and winter runoff predictions. With the decreasing of precipitation, the runoff prediction showed decreasing, while the corresponding 95% CI was little changed. With the increasing of temperature, both the spring runoff prediction and its 95% CI showed increasing.
展开▼
机译：水文模拟不确定性研究是当今水文科学研究中的热点问题。本文以流域水文模型SWAT为例，以黑河流域作为研究区，基于贝叶斯理论和方法，探讨SWAT模型参数不确定性对流量模拟和预测结果的影响。结果表明，在黑河流域，降水的变化幅度越大，模型参数不确定性对流量模拟和预测的影响越小;在降水变化相同条件下，温度增加会使模型参数不确定性对流量预测值的影响增大;在模型参数不确定性影响下，预测流量的不确定性区间夏季(尤其是6、7月份)最大;随着降水的减少预测流量值趋于减少，但预测流量值的不确定性区间变化不大;随着气温的升高，春季流量预测值有所增加，且春季流量预测值的不确定性区间也趋于增大;降水增加条件下，随着气温的降低，预测流量过程线越来越尖耸;降水减少条件下，随着气温的降低，预测流量过程线形状变化不大。 Uncertainty issue in hydrological modeling is a hot topic in recent hydrological research. Taking Heihe river basin as the study area, we mainly focused on the uncertainty in runoff prediction resulted from SWAT model parameter uncertainty. Bayesian method was employed for nine sensitive parameter estimations. We used 95% CI of runoff prediction to illustrate the uncertainty in runoff prediction caused by model parameter uncertainties. Results showed that: the larger the range of changing in precipitation, the narrower the 95% CIs of runoff prediction, the less effects of model parameter uncertainties to runoff prediction. With the increasing of temperature, the 95% CIs of runoff prediction were stretched if the precipitation kept stable, which means that the increases in temperature would lead to larger effects of model parameter uncertainty to runoff prediction. The 95% CIs of summer runoff prediction were the largest, followed by those of autumn and winter runoff predictions. With the decreasing of precipitation, the runoff prediction showed decreasing, while the corresponding 95% CI was little changed. With the increasing of temperature, both the spring runoff prediction and its 95% CI showed increasing.
展开▼
