摘要
局地微地形产生的微气象环境是造成气象预报误差的重要因素之一,也是导致覆冰预报准确性不高的主要原因。该研究利用高精度MODIS系统15 s(约500 m)地形数据驱动中尺度天气研究和预报(weather research and forecasting,WRF)模式,并使用基于决策树的梯度提升框架(light gradient boosting machine,LightGBM)对WRF预报进行订正,通过局地个例评估订正后的覆冰预测效果。结果表明:在假设条件下,过冷液滴覆冰速率随温度降低先快速增加,后增长速率保持不变,且液滴粒径越大,完全冻结所需温度越低;WRF-LightGBM订正算法在山区微地形下有效提升了温度预报准确度,典型冬季寒潮条件下预测温度与实际温度的误差在2℃以内,预报准确率为76%;以典型区域杆塔覆冰为例,输入订正后的温度和相对湿度数据后,覆冰融化时段被消除,覆冰厚度曲线与实际基本一致,增长速率接近一致。
The micro meteorological environment generated by local micro topography is one of the important factors causing meteorological forecasting errors,and also an important factor leading to the low accuracy of ice cover forecasting by power grid companies.Using high-precision MODIS system 15s(about 500m)terrain data to drive a mesoscale weather research and forecasting(WRF)model,and using light gradient boosting machine(LightGBM)to correct the WRF forecast,the improvement effect of ice cover prediction after correction is evaluated through local case studies.The results show that under the assumed conditions,the ice deposition rate of supercooled droplets increases rapidly with decreasing temperature,and then the growth rate remains unchanged.Moreover,the larger the droplet size,the lower the temperature required for complete freezing;The WRF LightGBM correction algorithm effectively improves the accuracy of temperature forecasting in mountainous micro terrain,with an error of less than 2℃between predicted and actual temperatures under typical winter cold wave conditions,and a forecasting accuracy of 76%;In a case of ice cover on a tower,after inputting corrected temperature and relative humidity data,the melting period of the ice cover was eliminated,and the ice cover thickness curve was basically consistent with the actual situation,with a growth rate close to the same.
作者
范强
肖书舟
张厚荣
叶华洋
付鑫怡
吴建蓉
FAN Qiang;XIAO Shuzhou;ZHANG Hourong;YE Huayang;FU Xinyi;WU Jianrong(Electric Power Research Institute of Guizhou Power Grid Co.,Ltd.,Guiyang 550002,Guizhou,China;Key Laboratory of Ice Prevention&Disater Reducing of China Southern Power Grid Co.,Ltd.,Guiyang 550002,Guizhou,China;Southern Power Grid Scientific Research Institute Co.,Ltd.,Guangzhou 510700,Guangdong,China)
出处
《电力大数据》
2024年第7期54-61,共8页
Power Systems and Big Data
基金
南方电网有限责任公司科技项目(GZKJXM20222326)。
关键词
精细化预报
微地形
数值预报
LightGBM
导线覆冰
覆冰增长速率
refined forecasting
microtopography
numerical forecasting
light gradient boosting machine
wire icing
ice cover growth rate
作者简介
范强(1986),男,硕士,高级工程师,主要从事生产技术管理及防灾减灾工作。
