| 数字滤波方法在月尺度数值预报中的应用Ⅰ.初始场滤波的作用 |
| |
| 引用本文： | 杨燕,纪立人.数字滤波方法在月尺度数值预报中的应用Ⅰ.初始场滤波的作用[J].大气科学,1997(4). |
| |
| 作者姓名： | 杨燕 纪立人 |
| |
| 作者单位： | 中国科学院大气物理研究所 |
| |
| 摘 要： | 本工作将数字滤波方法用于Ｔ４２Ｌ９谱模式的月预报实验。为了去除高频扰动误差的影响，保留对月预报有意义且可预报性较强的低频过程，用滤波器对观测资料的序列进行处理，得到１０天以上的低频分量作为初始场进行预报。结果说明，与不进行滤波的控制实验相比，误差增长明显减慢，预报效果在１０～２０天时段内提高最显著。进一步分析表明，延伸预报效果随时间的变化主要是由其中低频分量的预报效果决定的。初始场滤波后预报效果的提高也主要是由于其低频分量报得更好，而在此低频背景下新产生出来的高频扰动误差也相应较小，因而总体的预报效果好于控制实验。而如果在控制实验预报后提取低频分量进行检验，则对逐日预报的改进不明显。经过初始场滤波后对１～１０天至１～３０天平均场的预报也有较显著的提高。对旬平均预报改进最大的是在第二个１０天。对５００ｈＰａ环流形势的预报也更接近实际。本工作的结果说明，在中长期预报中如果能用适当的方法提取出低频过程的信息，则可望提高预报水平
|
| 关 键 词： | 数字滤波 数值预报 延伸预报 |
The Application of the Digital Filtering Method in Monthly Numerical Weather Prediction Part I: The Impact of Filtering the Initial Field |
| |
| Yang Yan,and,Ji Liren.The Application of the Digital Filtering Method in Monthly Numerical Weather Prediction Part I: The Impact of Filtering the Initial Field[J].Chinese Journal of Atmospheric Sciences,1997(4). |
| |
| Authors: | Yang Yan and Ji Liren |
| |
| Abstract: | This work is an experiment on the application of the digital filtering method in monthly extended-range forecast with T42L9 global spectral model. In order to eliminate the effect of fast growing errors of high-frequency (HF) perturbations and reserve the low-frequency (LF) processes, which are of significance in long-range forecast and which have relatively higher predictabilities, the observational data series is subject to a time filter to get the LF components with periods longer than 10 days. They are used as the initial conditions for the 30-day forecast (filtered run). The result shows that compared with the control run with no filtering, the error growth of the filtered run is obviously slower. The most significant improvement of forecast skill is in the period of 10 to 20 days. Further analyses show that the variation with time of the forecast skill in extended range is mainly determined by that of the LF modes. The improvement of forecast skill by the filtered run is due to its better skill in predicting LF modes than in the control run, and on its better LF background. The error of the newly-generated HF modes tends to be smaller. Thus the filtered run forecast is more accurate than the control run. On the other hand, if the LF components are extracted from the control run, the improvement on day-to-day forecast is not significant. The forecast skill for 1-5 day, 1-10 day to 1-30 day mean field is also significantly higher in the filtered run than in the control run. The most significant improvement is for the second 10 day mean. Also, the mean 500 hPa circulation pattern in the filtered run resembles the observation better than that in the control run. These results suggest that in extended NWP, if information can be extracted by LF processes properly, significant improvenents in forecast skill can be expected. |
| |
| Keywords: | digital filtering numerical weather prediction extended-range forecast |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《大气科学》浏览原始摘要信息 |
|
点击此处可从《大气科学》下载全文 |
|
