[1]黄小燕,韦春霞,赵华生,等.地面-雷达-卫星资料的广西降水临近预报应用效果评估[J].气象研究与应用,2022,43(04):50-58.[doi:10.19849/j.cnki.CN45-1356/P.2022.4.09]
　Huang Xiaoyan,Wei Chunxia,Zhao Huasheng,et al.Evaluation of the application effect of Ground-Radar-Satellite data in Guangxi precipitation proximity forecast[J].Journal of Meteorological Research and Application,2022,43(04):50-58.[doi:10.19849/j.cnki.CN45-1356/P.2022.4.09]
点击复制

地面-雷达-卫星资料的广西降水临近预报应用效果评估()
分享到：

气象研究与应用[ISSN:1673-8411/CN:45-1356/P]

卷:
第43卷
期数:
2022年04期
页码:
50-58
栏目:
研究论文
出版日期:
2022-12-15

文章信息/Info

Title:
Evaluation of the application effect of Ground-Radar-Satellite data in Guangxi precipitation proximity forecast
作者:
黄小燕1; 2;  韦春霞1;  赵华生1;  何立1;  吴玉霜3; 4
1. 广西壮族自治区气象科学研究所, 南宁 530022;
2. 防城港暴雨实验室, 广西 防城港 538001;
3. 广西壮族自治区气象台, 南宁 530022;
4. 广西壮族自治区气象灾害防御技术中心, 南宁 530022
Author(s):
Huang Xiaoyan1; 2;  Wei Chunxia1;  Zhao Huasheng1;  He Li1;  Wu Yushuang3; 4
1. Guangxi Institute of Meteorological Sciences, Nanning 530022, China;
2. Fangchenggang Rainstorm Laboratory, Guangxi Fangchanggang 538001, China;
3. Guangxi Meteorological Observatory, Nanning 530022, China;
4. Guangxi Meteorological Disaster P
关键词:
地面实况观测数据; 雷达组合反射率; 风云四号A星波段数据; 降水临近预报; 评估
Keywords:
Ground observation data; radar combined reflectivity; band data of FY4A satellite; precipitation approaching forecast; evaluate
分类号:
P457
DOI:
10.19849/j.cnki.CN45-1356/P.2022.4.09
摘要:
基于地面实况观测数据、雷达组合反射率、风云四号A星波段数据3种实况观测资料,以2018-2021年广西2850个站点的小时累计雨量为预报对象,采用随机森林算法建立未来1-3h的降水临近预报模型,并分别进行了单类型观测资料预报因子、3种类型观测资料多种组合预报因子输入的预报试验。对各预报试验结果综合采用TS评分、命中率、虚警率和漏报率进行点对点的评估表明,地面资料在1-3h的小雨和中雨量级、1-2h大雨量级的预报能力较好;雷达资料对未来1h暴雨量级的预报较其他两种观测资料优势明显;卫星资料在小雨量级上有一定的预报能力,但其他量级各时效的预报均不理想。3种观测资料在第2-3h的暴雨预报能力都偏低。3类预报资料因子组合预报结果的评估表明,各量级的多数预报时效均能比单类型资料预报取得更高的预报精度,其中大雨量级的预报提高了10%以上,暴雨则提高了25%以上。大雨和暴雨1hTS评分的空间分布情况表明,组合因子高评分区域分布均最广,地面和雷达大雨量级的空间分布相当,雷达的暴雨TS评分在0.2以上空间分布范围较地面和卫星资料广,卫星的TS评分的空间分布显示其预报能力均最弱。
Abstract:
Based on three kinds of ground observation data, radar combined reflectivity and band data of FY4A satellite from the live observation data, taking the hourly accumulated rainfall of 2850 stations in Guangxi from 2018 to 2021 as the prediction object, the random forest algorithm is used to establish the rainfall approaching prediction model for the next 1-3 hours, and the prediction tests of single type observation data prediction factor and three types of observation data multiple combination prediction factor input are conducted respectively. The results of each prediction test were evaluated by TS score, hit rate, false alarm rate and false alarm rate, the ground data has a good prediction ability in light rain and moderate rain in the next 1-3 hours and heavy rain in the next 1-2 hours; the radar data has obvious advantages over the other two kinds of observation data in forecasting the rainstorm magnitude in the next hour; the satellite data has a certain forecast ability in the light rain magnitude, but the predictions of other magnitudes and time effects are not ideal. The rainstorm forecast capacity of all three observations was low at the 2nd-3rd hour. The evaluation of the combined forecast results of three types of forecast data factors shows that the prediction accuracy of most forecasts of all magnitudes is higher than that of single-type data. The prediction of heavy rain increases by more than 10%, and that of heavy rain increases by more than 25%. The spatial distribution of 1-hour TS scores of heavy rain and rainstorm shows that the regions with high combined factor scores are the most widely distributed, and the spatial distribution of ground and radar heavy rain magnitude is similar. The spatial distribution range of radar heavy rain TS scores above 0.2 is wider than that of ground and satellite data, and the spatial distribution of satellite TS scores shows that their forecasting ability is the weakest.

参考文献/References:

[1] 俞小鼎,周小刚,王秀明. 雷暴与强对流临近天气预报技术进展[J]. 气象学报,2012,70(3):311-337.
[2] 范水勇,王洪利,陈敏,等.雷达反射率资料的三维变分同化研究[J].气象学报,2013,71(3):527-537.
[3] 郑淋淋,邱学兴. 一种改进的降水临近外推预报技术方法研究及效果检验[J].气象科技,2020,48(1):97-106.
[4] 沈文海. 再析气象大数据及其应用[J]. 中国信息化,2016(1):84-96.
[5] 韩雷,王洪庆,林隐静. 光流法在强对流天气临近预报中的应用[J]. 北京大学学报(自然科学版),2008,44(5):751-755.
[6] 韩丰,魏鸣,李南,等. 反射率因子和径向风速共同约束反演多普勒雷达风场化[J]. 遥感学报,2013,17(3):584-589.
[7] 韩丰,龙明盛,李月安,等. 循环神经网络在雷达临近预报中的应用[J]. 应用气象学报,2019,30(1):61-69.
[8] 曹春燕,陈元昭,刘东华,等. 光流法及其在临近预报中的应用[J]. 气象学报,2015,73(3):471-480.
[9] Bechini R,Chandrasekar V. An enhanced optical flow technique for radar nowcasting of precipitation and winds[J]. Journal of Atmospheric and Ocean Technology,2017,34(12):2637-2658.
[10] Chen Y Z,Lan H P,Chen X L,et al. A nowcasting techniquebased on application of the particle filter blending algorithm[J]. Journal of Meteorological Research,2017,31(5):931-945.
[11] 梁振清,陈生.基于深度学习和雷达观测的华南短临预报精度评估[J].气象研究与应用,2020,41(1):41-47.
[12] 张佳洛,黄勇,刘传才. 基于卷积门循环单元和气象雷达图像的临近降水预报[J]. 计算机与数字工程,2021,49(8):1538-1542.
[13] 曹伟华,南刚强,陈明轩,等. 基于深度学习的京津冀地区精细尺度降水临近预报研究[J]. 气象学报,2022,80(4):546-564.
[14] 师春香,谢正辉. 基于静止气象卫星观测的降水时间降尺度研究[J]. 地理科学进展,2008,27(4):15-22.
[15] 魏建苏,严明良,樊永富,等. 卫星云图和数值产品结合的讯期强降水预警系统[J]. 气象科学,2001,21(3):355-362.
[16] John R, Mecikalsk, Kristopher M,et al. Forecasting convective initiation by monitoring the evolution of moving cumulus in daytime GOES imagery[J]. MECIKALSKI AND BEDKA,2006(1):49-77.
[17] 卢乃锰,吴蓉璋. 强对流降水云团的云图特征分析[J]. 应用气象学报,1997(3):14-20.
[18] 李嘉睿,卢乃锰,谷松岩. 青藏高原地区TRMM PR地面降雨率的修正[J]. 应用气象学报,2015,(5):636-640.
[19] 罗艳艳,吴雪贞,郑宝智,等. 基于区域自动站资料的闽中地区前汛期短时强降水特征分析[J]. 海峡科学,2016(9):7-10.
[20] 张功文,董方亮,谢详永. 基于区域自动站资料的邯郸市夏季短时强降水分布研究[J]. 现代农业科技,2016(14):205-207,220.
[21] 石宏辉,顾欣,杨通荣,等. 利用区域自动站资料作短时临近预报预警[J]. 贵州气象,2010,34(1):29-30.
[22] 丁立善,张长栋. 利用区域自动站资料作短时临近预报预警[J]. 农业与技术,2018,38(8):230.
[23] Breiman L. Random forests[J]. Machine Learning, 2001,45(1):5-32.

备注/Memo

备注/Memo:
收稿日期:2022-11-02。
基金项目:广西重点研发计划项目(桂科AB22035016和桂科AB21196041)、广西气象局重点基金项目(桂气科2021Z02)、国家自然科学基金项目(41765002)、广西人工智能预报技术创新团队项目
作者简介:黄小燕(1978-),女,博士,正研级高级工程师,从事智能计算预报技术在天气预报中的研究应用。E-mail:Gx_huangxy@163.com
通讯作者:韦春霞(1968-),女,本科,高级工程师,从事气象应用研究工作。E-mail:wcx_hc@163.com
更新日期/Last Update: 1900-01-01