To investigate the influence of different initial variables on the uncertainty of precipitation prediction, initial perturbations were generated in AREM model analysis field by scale decomposition based on spatial smoothing. As these perturbations were imposed upon the model initial analysis, several sets of sensitivity experiments were studied. The results show that heavy-rain prediction is more sensitive to the initial relative humidity, the temperature and the wind than to the geo-potential height, and that among the above 4 variables, the geo-potential height has the lowest error growth rate, which indicates that the influence of geo-potential height on the prediction precipitation is marginal. Since these variables are correlated with each other, the improvement of heavy-rain prediction is intrinsically limited if any of them is changed separately.%为了研究初始场不确定性对暴雨预报的影响,利用尺度分离的方法,在AREM模式分析场上构造了初值扰动,并通过在模式初始物理量上叠加或扣除这些扰动,设计了不同的敏感性试验.试验结果表明,暴雨预报结果对模式的初始相对湿度、温度和风场是敏感的,而对初始位势高度的敏感性较弱.由于各物理量是相互联系和制约的,单独改变某个物理量对暴雨预报精度的提高是有限的.
展开▼
