熊敏诠. 2017. 基于集合预报系统的日最高和最低温度预报[J]. 气象学报, ():-, doi:10.11676/qxxb2017.023
基于集合预报系统的日最高和最低温度预报
XIONG Minquan
投稿时间:2016-08-25  最后修改时间:2016-12-06
DOI:10.11676/qxxb2017.023
中文关键词:  集合预报系统  日最高温度 日最低温度 BP-SM方法
英文关键词:Ensemble  prediction system, Daily  maximum temperature, Daily  minimum temperature  BP neural  network - Self  Memory method
基金项目:
作者单位E-mail
熊敏诠 系统实验室 minquanxiong@sina.com 
摘要点击次数: 611
全文下载次数: 258
中文摘要:
      根据欧洲中心集合预报系统2米温度预报的集合统计值,提出了BP-SM方法,针对我国512个台站2016年3月的日最高(低)温度作预报分析。将集合预报系统的模式直接输出、BP和BP-SM方法得到的日最高(低)温度进行了比较,结果表明:当预报时效越长,BP-SM方法较之BP方法的预报优势也更明显;在1至5天的预报中,BP-SM方法显著降低了预报绝对误差,误差在2℃以内的准确率大部在60%~70%以上,部分站点达到了90%;正技巧评分均值大多高于30%,在青藏高原东部和南部地区超过了60%。预报正技巧站点次数在绝对误差≤2℃(1℃)范围内有所提高,对日最高温度预报准确率的提高略好于日最低温度;BP-SM方法有效地降低了预报系统偏差,较大预报误差出现次数显著减少。
英文摘要:
      BP neural network - Self Memory method (BP-SM) is used to calibrate Daily 2-m maximum (minimum) temperature forecasts over 512 stations of China with the European Centre for Medium-Range Weather Forecasts (ECMWF) Ensemble Prediction System (EPS) in March 2016. Seven statistical characteristics as predictor are calculated based on 2-m temperature model output in EPS. Daily maximum (minimum) temperature forecasts by BP-SM and by direct model output (DMO) are compared. The postprocessing with BP-SM is shown to improve the forecast accuracy. The accurate rate of Daily maximum (minimum) temperature forecasts absolute error which is less than 2℃ reaches above 60% or over 90% . The forecasting skill score relative to BP-SM is 30% averagely, and above 60% over Eastern Tibet, compared with DMO. This program is obviously superior within 2℃(1℃). The calibrated daily 2-m maximum temperature is slightly better than the daily 2-m minimum temperature. By BP-SM, the systematic deficiencies of daily maximum (minimum) temperature forecasts significantly reduced.
查看全文   查看/发表评论  下载PDF阅读器
分享按钮