为充分了解西北地区的高分辨率快速循环同化预报系统(Rapid Update Cycle, RUC)的预报性能, 对该系统2012年6月2日至7月6日的预报结果进行了检验和评估。利用双线性插值方法将西北RUC系统的预报结果插值到最近的观测站点上, 计算500 hPa和700 hPa位势高度场、 温度场、 风向以及风速的平均误差和均方根误差, 对预报结果进行检验评估; 24 h定量降水利用累计降水分级检验方法, 检验的统计量包括TS评分、 偏差(BIA)、 公平T评分(ETS)和真实技巧评分(TSS)。检验分析表明: (1)500 hPa的高度场预报、 温度场预报、 风速预报和地面2 m温度预报都存在着正的系统性偏差, 其24 h平均误差均值分别为0.17 gpm、 0.63 ℃、 1.19 m·s-1和1.49 ℃。700 hPa高度场和温度场预报存在着负的系统性偏差, 其24 h平均误差均值分别为-0.41 gpm和 -0.11 ℃。 (2)除了风速、 风向, 其他要素24 h预报结果的均方根误差均值都小于48 h预报结果的均方根误差均值, 500 hPa高度场和温度场的24 h预报的均方根误差均值分别为1.32 gpm和1.37 ℃, 而其48 h值则分别为1.56 gpm和1.53 ℃; 700 hPa高度场和温度场的24 h预报的均方根误差均值分别为1.21 gpm和1.40 ℃, 而其48 h值则分别为1.38 gpm和1.94 ℃; 2 m温度的24 h和48 h的均方根误差均值分别为3.06 ℃和3.30 ℃, 表明随着预报时效的增加, 预报性能降低, 这与模式预报性能相符。 (3)24 h定量降水分级的TS评分、 ETS评分和TSS评分几乎都有相同的大值中心, 说明模式对于这些大值中心附近地区的各量级降水预报效果比较好。总体上, 模式对于大雨和暴雨预报效果较好的地区处于西北地区东南部。从偏差(BIA)评分来看, 模式对青海南部与四川北部交界地区的降水预报并不理想, 表现为小雨预报漏报较多, 而对该地区中雨和大雨预报空报较多。
To explore the operational performance of northwest high?resolution Rapid Update Cycle (RUC) system, the operational forecasts during 2nd June to 6th July of 2012 were verified and evaluated.The Northwest RUC predictive results were interpolated to the nearby observation site using the bilinear interpolation method.Statistical methods were used to calculate mean error (ME) and root mean square error (RMSE) of geopotential height field, temperature, wind direction and wind speed at 500 hPa and 700 hPa, then the score of TS, deviation (BIA), T score (ETS) fair and true skill score (TSS) were calculated to evaluate the quantitative precipitation predicted in 24 hours.The results are as follows: (1) there were systematic positive deviation for height field prediction, temperature field prediction, and wind speed predication at 500 hPa and temperature at 2?meter height, the 24?hour average of ME were 0.17 gpm, 0.63 ℃, 1.19 m·s-1, 1.49 ℃, respectively.Height field prediction and temperature field prediction at 700 hPa showed systematic negative deviation, the 24?hour average of ME was -0.41 gpm, -0.11 ℃, respectively.(2) Except for wind speed and wind direction, other elements forecast effect in 24?hour were better than the forecast effect in 48?hour of the model, which match to the prediction performance.The 24?hour average of RMSE of geopotential height field and temperature at 500 hPa were 1.32 gpm, 1.37 ℃, respectively, and the value of 48?hour of RMSE were 1.56 gpm, 1.53 ℃, respectively.The 24?hour average of RMSE of geopotential height field and temperature at 700 hPa was 1.21 gpm, 1.40 ℃, respectively, and the value of 48?hour of RMSE were 1.38 gpm, 1.94 ℃, respectively.The 24?hour and 48?hour average of RMSE of temperature at 2?meter were 3.06 ℃, 3.30 ℃, respectively.(3) The TS score, ETS score and TSS score of quantitative precipitation were almost have the same large value center at different precipitation grade in 24 hours.It means the precipitation forecast effect was comparatively good near the center region of the large value.The heavy rain and rainstorm forecast effect was better in the southeastern of Northwest area.From the BIA score, the model forecasted less light rain events than observed, but forecasted more moderate rain and heavy rain events than observed in the south of Qinghai and north of Sichuan.
[1]Atencia A, Rigo T, Sairouni A, al et, 2010.Improving QPF by blending techniques at the Meteorological Service of Catalonia[J].Natural Hazards and Earth System Science, 10(7): 1443 -1455.DOI: 10.5194/nhess-10-1443-2010.
[2]Auligne T, Menetrier B, Lorenc, A C, al et, 2016.Ensemble?variational integrated localized data assimilation[J].Monthly Weather Review, 144(10): 3677-3696.DOI: 10.1175/MWR-D-15-0252.1.
[3]Benjamin S G, Devenyi D, Weygandt S S, al et, 2004.An assimilation?forecast cycle: The RUC[J].Monthly Weather Review, 132: 495-518.
[4]Golding B W, 1998.A system for generating automated very short range forecasts[J].Meteorological Applications, 5(1): 1-16.
[5]Payne T J, 2017.Rapid update cycling with delayed observations[J].Tellus A: Dynamic Meteorology and Oceanography, 69 (1):1409061.DOI: 10.1080/16000870.2017.1409061.
[6]陈敏, 范水勇, 郑祚芳, 等, 2011.基于BJ?RUC系统的临近探空及其对强对流发生潜势预报的指示性能初探[J].气象学报, 69(1): 181-194.DOI: 10.11676/qxxb2011.016.
[7]陈敏, 范水勇, 仲跻芹, 等, 2010.全球定位系统的可降水量资料在北京地区快速更新循环系统中的同化试验[J].气象学报, 68(4): 450-463.DOI: 10.11676/qxxb2010.044.
[8]范水勇, 陈敏, 仲跻芹, 等, 2009.北京地区高分辨率快速循环同化预报系统性能检验和评估[J].暴雨灾害, 28(2): 119-125.
[9]范水勇, 郭永润, 陈敏, 等, 2008.高分辨率WRF三维变分同化在北京地区降水预报中的应用[J].高原气象, 27(6): 1181-1188.
[10]郝民, 徐枝芳, 陶士伟, 等, 2011.GRAPES RUC系统模拟研究及应用试验[J].高原气象, 30(6): 1573-1583.
[11]胡胜, 罗聪, 黄晓梅, 等, 2012.基于雷达外推和中尺度数值模式的定量降水预报的对比分析[J].气象, 38(3): 274-280.DOI: 10.7519/j.issn.1000-0526.2012.3.003.
[12]黄玉霞, 王宝鉴, 王研峰, 等, 2017.甘肃省夏季暴雨日数特征及其与大气环流关系[J].高原气象, 36(1): 183-194.DOI: 10. 7522/j.issn.1000-0534.2015.00118.
[13]雷蕾, 孙继松, 王国荣, 等, 2012.基于中尺度数值模式快速循环系统的强对流天气分类概率预报试验[J].气象学报, 70(4): 752-765.DOI: 10.11676/qxxb2012.061.
[14]盛春岩, 薛德强, 雷霆, 等, 2006.雷达资料同化与提高模式水平分辨率对短时预报影响的数值对比试验[J] .气象学报, 64(3): 293-308.DOI: 10.11676/qxxb2006.028.
[15]王丹, 王改利, 刘黎平, 等, 2014.基于雷达回波外推和中尺度模式预报的短时降水对比分析[J].高原气象, 33(3): 811-822.DOI: 10.7522 /j.issn.1000-0534.2013.00038.
[16]魏东, 尤凤春, 杨波, 等, 2011.北京快速更新循环预报系统(BJ?RUC)要素预报质量评估[J].气象, 37(12): 1489-1497.
[17]魏东, 尤凤春, 范水勇, 等, 2010.北京快速更新循环预报系统(BJ?RUC)探空预报质量评估分析[J], 气象, 36(8): 72-80.DOI: 10.7519/j.issn.1000-0526.2010.8.010.
[18]熊秋芬, 2011.GRAPES_Meso模式的降水格点检验和站点检验分析[J].气象, 37(2): 185-193.DOI: 10.7519/j.issn.1000-0526.2011.2.008.
[19]杨显玉, 文军, 王大勇, 等, 2016.一次甘肃强降水过程的数值模拟与诊断分析[J].高原气象, 35(3): 726-733.DOI: 10.7522/j.issn.1000-0534.2015.00048.
[20]尤凤春, 魏东, 王雨, 2009.北京奥运期间多模式降水检验及集成试验[J].气象, 35(11): 3-8.DOI: 10.7519/j.issn.1000-0526.2009.11.001.
[21]张守峰, 郭文华, 张金艳, 等, 2003.T213模式对强冷空气短期预报能力的分析检验[J].气象, 29(8): 43-47.DOI: 10.7519/j.issn.1000-0526.2003.8.010.