一次长江流域梅雨降水中三种云量计算方案的对比研究

Comparison of the performance of three schemes in calculating the cloudiness during a Mei-yu rainfall process over the Yangtze River Basin

采用NCEP分析场,选取2010年梅雨期长江流域的一次降水过程,分别基于Slingo方案、NCAR方案和钱氏方案,利用相对湿度计算云量,并以LAPS(Local Analysis and Prediction System)系统输出的云量分析场作为观测值,分别在高层(400 h Pa)与低层(850 h Pa),从宏观比较与统计分析的角度,与计算结果进行云量大小与区域分布的对比分析。结果表明,三个云量计算方案对云量中心位置的把握均较为准确,但对云量值的计算存在大小不等的误差。NCAR方案计算结果和LAPS输出场最为吻合,能够体现出云量大值区,但区域一般偏大;Slingo方案相较NCAR方案来说略差,但也能较好地描述云带分布;此外,钱氏方案计算出的云量值始终偏小,但其能够较好地描述云带轮廓与云量的分布特征。综合对比结果,NCAR云量计算方案比其余两者更优,且在低层(850 h Pa)表现尤为明显。

Cloud is an important internal factor of the climate system,especially in the earth—atmosphere system. The formation of clouds and their characteristics mainly result from both dynamic and thermodynamic processes of the surface and the atmosphere. An accurate grasp of the distribution of cloud and its variability can contribute greatly when attempting to assess the role of cloud in climate change. And related to this,a calculation scheme that is effective at describing cloudiness is a key part of improving the ability to simulate cloud in numerical models. In order to identify a satisfactory cloudiness calculation scheme,the present study employed NCEP reanalysis data to calculate cloudiness and relative humidity,based on three schemes（ Slingo,NCAR,and Qian）,during a Mei-yu rainfall process in the Yangtze River basin in 2010. Comparative analysis between the calculation results and Local Analysis and Prediction System（ LAPS） reanalysis data,considered as the cloud observation,was conducted in terms of cloud distribution and cloudiness values at upper and lower levels,represented by 400 h Pa and 850 h Pa,respectively. Based on the comparison results,as well as statistical analysis involving anomaly correlation coefficients（ ACCs） and RMSE,it was found that the three cloudiness calculation schemes all managed to successfully simulate the cloud central positions,but each had their own advantages and particular characteristics when it came to cloudiness values. The results calculated using the NCAR scheme matched the LAPS outputs very well at largevalue centers of cloud,but the regions containing these values were always too large. To a certain extent,although it was found that the Slingo scheme could also describe the cloud well,it showed a slightly lower capacity than the NCAR scheme in terms of its cloudiness calculation. Additionally,the Qian scheme demonstrated fairly limited ability to calculate the cloudiness values,but always presented the cloud profile and its distribution accurately.Based on the statistical analysis,at 850 h Pa,the NCAR scheme produced its maximum ACC and minimum RMSE,indicating its superiority over the other schemes at calculating the cloudiness at that height. However,the Qian scheme yielded the best statistical results at 400 h Pa,possibly due to the close correspondence of its cloudiness distribution results with the observation. Of importance here is that,to a certain extent,the Qian scheme takes into account the cloud formation mechanism and the influence of atmospheric vertical motion on cloud formation when calculating the cloudiness. Overall,based on this comprehensive comparison of relevant factors,we conclude that the NCAR scheme is superior to the others,particularly at the lower level（ 850 h Pa）.