| 摘要: |
| 利用2018年1～10月华南3Km区域高分辨率模式08:00、20:00起报的气温预报和实况资料，采用线性内插法进行站点预报值处理，并从平均均方根误差及预报准确率的角度，检验分析了贵州省72h预报内逐24h最高（低）气温预报质量。结果表明，72h内随着预报时效的增加，预报准确率差异较小；日最低气温预报准确率相对最高气温平均高出20%左右；08:00起报的最高（低）气温预报优于20:00的。同时发现，最高（低）气温的预报能力在月份上存在明显差异，6～8月预报性能总体优于其它月份；在24～48h预报中，东北-西南向一带较贵州其它区域展现出更高的预报能力。在9个主要城市站上，最高（低）气温均表现出较高的预报技巧，其中，20:00起报的兴义站24h最低气温准确率100%。通过对2018年7月18日气温预报质量检验，最高（低）气温及35.0℃以上高温事件预报准确率均在80%左右，较好反映了天气实况。因此，华南3Km高分辨率区域模式对贵州气温预报具有较好的参考价值。 |
| 关键词: 华南；高分辨率；气温；检验 |
| DOI： |
| 投稿时间：2019-05-28修订日期：2019-08-13 |
| 基金项目:气象预报业务关键技术发展专项(YBGJXM（2019）03-07)：复杂山地背景下网格要素预报订正技术研发; 国家预报员专项（CMAYBY2016-065）：深秋初冬时节静止锋减弱北抬对贵州气温的差异性分析；2019年贵州省气象台业务项目：贵州省实况格点偏差订正及研究。 |
|
| Quality test of 3km high resolution regional model temperature prediction in Guizhou province |
| LI Gang,PENG Fen,KONG Dexuan,LIU Yanhua,WU Lei,LI Yanlin |
| (Guizhou Meteorological Observatory;Guizhou Institute on Mountainous Environment and Climate;Meteorological Office of southwestern Guizhou) |
| Abstract: |
| According the comparison between observation temperature from station and 3Km meso-scale model for south China temperature forecast starting at 00:00 GMT and 12:00 GMT from January to October of 2018 over Guizhou. By using bilinear interpolation method, the characteristics of 24-72h prediction errors of daily maximum（minimum）are analyzed. The result show that the prediction performance rarely change with increase forecast time, also the forecast accuracy scores of minimum temperature exceeds 20% for the maximum temperature, also accuracy of the predicted temperature starting at 00:00 GMT is better than 12:00 GMT. In addition, there is obvious month‘s variation, higher in June to August compared with other months. Meanwhile the northeast-southwest area showed higher forecasting skills than other areas in the 24-48h. In each observation station, the maximum (minimum) temperatures shows higher forecasting skills, also the prediction accuracy of 24h minimum temperature at Xingyi station reached 100% at 12:00 GMT. Through the test of the case July 18, 2018, the maximum（minimum）temperature and high temperature events forecast accuracy are around 80%。Compared to local temperature forecast, the meso-scale model are of more reliable reference significance. |
| Key words: South China;high resolution;temperature;verification |
