Chen Yuwen, Huang Xiaomeng, Li Yi, et al. Ensemble learning for bias correction of station temperature forecast based on ECMWF products. J Appl Meteor Sci, 2020, 31(4): 494-503. DOI:  10.11898/1001-7313.20200411.
Citation: Chen Yuwen, Huang Xiaomeng, Li Yi, et al. Ensemble learning for bias correction of station temperature forecast based on ECMWF products. J Appl Meteor Sci, 2020, 31(4): 494-503. DOI:  10.11898/1001-7313.20200411.

Ensemble Learning for Bias Correction of Station Temperature Forecast Based on ECMWF Products

DOI: 10.11898/1001-7313.20200411
  • Received Date: 2020-01-15
  • Rev Recd Date: 2020-04-09
  • Publish Date: 2020-07-31
  • To improve the accuracy of numerical weather prediction (NWP) and its ability for extreme weather event forecast, a hybrid model based on ensemble learning is proposed and tested by post-processing one of the most successfully predicted variables, temperature at 2 m height. The NWP dataset used is provided by The International Grand Global Ensemble (TIGGE) project in the European Centre from Medium-Range Weather Forecasts (ECMWF), with a horizontal resolution of 0.125°×0.125° and lead times from 6 to 168 h (with a 6 h increment, 28 lead times totally). The observation is collected from 301 stations covering China expect for Xizang and Qinghai, including 4 variables, temperature, pressure, relative humidity and wind speed every 3 hours. The ECMWF product and observation span a period of 6 years ranging from 1 January 2013 to 31 December 2018. Data from 2013 to 2017 are used for machine learning and model training, and data in 2018 are used for testing. The hybrid model named ALS consists of 2 stages. Stage 1 trains two separate models, a long short-term memory combined with a fully connected neural network (LSTM-FCN) and an artificial neural network (ANN). Stage 2 blends the output of LSTM-FCN and ANN using a linear regression (LR) model. The correction result is the output of LR. ALS model is then applied to correct the station temperature forecast with lead time from 6 to 168 h. Outcomes are verified by observations from stations, while LR model is used as control model. ALS model reduces the average root mean square error (RMSE) of the station temperature forecast by 0.61℃ (19.6%), and by 0.23℃ (8.4%) compared with the LR model. ALS model reduces RMSE at more stations compared with LR model (252 vs. 186). ALS model is particularly effective in areas where the accuracy of station temperature forecast is low, such as Guizhou and Yunnan. Forecasts for stations in these areas are significantly improved with an average RMSE reduction over 40%. Moreover, case analysis of high temperature show that ALS model improves the forecast accuracy of high temperature events significantly, with a RMSE reduction of 30.5% at 4 stations compared to station temperature forecast. It demonstrates that ensemble learning can be used to supplement weather forecast.
  • Fig. 1  The structure of ALS model

    Fig. 2  average correction improvement rate of different models compared to station temperature forecast

    (the blue denotes a positive correction improvement rate, the red denotes a negative correction improvement rate)

    Fig. 3  Comparison of temperature forecast with lead time of 72 h to the observation at Guiyang, Yuanping, Fuzhou and Tainan from Jun to Aug in 2018

    Fig. 4  Comparison of averaged root mean square error of temperature forecast at Guiyang, Yuanping, Fuzhou and Tainan from 15 Jul to 22 Jul in 2018

    Table  1  Root mean square error of temperature forecast with different lead times in 2018(unit:℃)

    DownLoad: Download CSV

    Table  2  Root mean square error of temperature forecast with lead time of 72 h at 4 stations from Jun to Aug in 2018(unit:℃)

    DownLoad: Download CSV
  • [1]
    Bauer P, Thorpe A, Brunet G.The quiet revolution of numerical weather prediction.Nature, 2015, 525(7567):47.
    Lynch P.The origins of computer weather prediction and climate modeling.Journal of Computational Physics, 2008, 227(7):3431-3444.
    Glahn H R, Lowry D A.The use of model output statistics (MOS) in objective weather forecasting.J Appl Meteor, 1972, 11(8):1203-1211. doi:  10.1175-1520-0450(1972)011-1203-TUOMOS-2.0.CO%3b2/
    薛谌彬, 陈娴, 张瑛, 等.ECMWF高分辨率模式2 m温度预报误差订正方法研究.气象, 2019, 45(6):831-842.
    张玉涛, 佟华, 孙健.一种偏差订正方法在平昌冬奥会气象预报的应用.应用气象学报, 2020, 31(1):27-41. doi:  10.11898/1001-7313.20200103
    吴启树, 韩美, 郭弘, 等.MOS温度预报中最优训练期方案.应用气象学报, 2016, 27(4):426-434. doi:  10.11898/1001-7313.20160405
    刘还珠, 赵声蓉, 陆志善, 等.国家气象中心气象要素的客观预报——MOS系统.应用气象学报, 2004, 15(2):181-191.
    Overpeck J T, Meehl G A, Bony S, et al.Climate data challenges in the 21st century.Science, 2011, 331(6018):700-702.
    Zurada J M.Introduction to Artificial Neural Systems.St Paul:West Publishing Company, 1992.
    Gers F A, Schmidhuber J, Cummins F.Learning to forget:Continual prediction with LSTM.Neural Computation, 12(10):2451-2471.
    Krizhevsky A, Sutskever I, Hinton G E.Imagenet Classification with Deep Convolutional Neural Networks//Advances in Neural Information Processing Systems, 2012: 1097-1105.
    Reichstein M, Camps-Valls G, Stevens B, et al.Deep learning and process understanding for data-driven earth system science.Nature, 2019, 566(7743):195.
    Scher S.Toward data-driven weather and climate forecasting:approximating a simple general circulation model with deep learning.Geophys Res Lett, 2018, 45(22):12616-12622. doi:  10.1029/2018GL080704
    Rasp S, Pritchard M S, Gentine P.Deep learning to represent subgrid processes in climate models.Proceedings of the National Academy of Sciences, 2018, 115(39):9684-9689.
    Ham Y G, Kim J H, Luo J J.Deep learning for multi-year ENSO forecasts.Nature, 2019, 573(7775):568-572.
    Pan B, Hsu K, AghaKouchak A, et al.Improving precipitation estimation using convolutional neural network.Water Resources Research, 2019, 55(3):2301-2321.
    王彦磊, 曹炳伟, 黄兵, 等.基于神经网络的单站雾预报试验.应用气象学报, 2010, 21(1):110-114.
    闵晶晶, 孙景荣, 刘还珠, 等.一种改进的BP算法及在降水预报中的应用.应用气象学报, 2010, 21(1):55-62.
    杨璐, 韩丰, 陈明轩, 等.基于支持向量机的雷暴大风识别方法.应用气象学报, 2018, 29(6):680-689. doi:  10.11898/1001-7313.20180604
    Runge J, Bathiany S, Bollt E, et al.Inferring causation from time series in earth system sciences.Nature Communications, 2019, 10(1):2553.
    Liu Y, Racah E, Correa J, et al.Application of Deep Convolutional Neural Networks for Detecting Extreme Weather in Climate Datasets.arXiv Preprint arXiv: 1605.01156, 2016.
    Polikar R.Ensemble Learning.Boston:Springer, 2012:1-34.
    Bougeault P, Toth Z, Bishop C, et al.The THORPEX interactive grand global ensemble.Bull Amer Meteor Soc, 2010, 91(8):1059-1072.
    Swinbank R, Kyouda M, Buchanan P, et al.The TIGGE project and its achievements.Bull Amer Meteor Soc, 2016, 97(1):49-67.
    Myers D E.Spatial interpolation:An overview.Geoderma, 1994, 62(1-3):17-28.
    高歌, 龚乐冰, 赵珊珊, 等.日降水量空间插值方法研究.应用气象学报, 2007, 18(5):732-736. doi:  10.11898/1001-7313.20070511
    彭彬, 周艳莲, 高苹, 等.气温插值中不同空间插值方法的适用性分析——以江苏省为例.地球信息科学学报, 2011, 13(4):539-548.
    潘留杰, 薛春芳, 王建鹏, 等.一个简单的格点温度预报订正方法.气象, 2017, 43(12):1584-1593.
    Karlik B, Olgac A V.Performance analysis of various activation functions in generalized MLP architectures of neural networks.International Journal of Artificial Intelligence and Expert Systems, 2011, 1(4):111-122.
    Nair V, Hinton G E.Rectified Linear Units Improve Restricted Boltzmann Machines//Proceedings of the 27th International Conference on Machine Learning (ICML-10), 2010: 807-814.
    Connor J T, Martin R D, Atlas L E.Recurrent neural networks and robust time series prediction.IEEE Transactions on Neural Networks, 1994, 5(2):240-254.
    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.应用气象学报, 2019, 30(1):61-69. doi:  10.11898/1001-7313.20190106
    Qing X, Niu Y.Hourly day-ahead solar irradiance prediction using weather forecasts by LSTM.Energy, 2018, 148:461-468.
    Cao Y, Gui L.Multi-step Wind Power Forecasting Model Using LSTM Networks, Similar Time Series and LightGBM//5th International Conference on Systems and Informatics (ICSAI).IEEE, 2018:192-197.
    Shi X, Chen Z, Wang H, et al.Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting//Advances in Neural Information Processing Systems, 2015: 802-810.
    Woźniak M, Graña M, Corchado E.A survey of multiple classifier systems as hybrid systems.Information Fusion, 2014, 16:3-17.
    Glahn H R, Lowry D A.The use of model output statistics (MOS) in objective weather forecasting.J Appl Meteor, 1972, 11(8):1203-1211. doi:  10.1175-1520-0450(1972)011-1203-TUOMOS-2.0.CO%3b2/
    曾晓青, 薛峰, 姚莉, 等.针对模式风场的格点预报订正方案对比.应用气象学报, 2019, 30(1):49-60. doi:  10.11898/1001-7313.20190105
    Marzban C, Sandgathe S, Kalnay E.MOS, perfect prog, and reanalysis.Mon Wea Rev, 2006, 134(2):657-663.
    Hart K A, Steenburgh W J, Onton D J, et al.An evaluation of mesoscale-model-based model output statistics (MOS) during the 2002 Olympic and Paralympic Winter Games.Wea Forecasting, 2004, 19(2):200-218.
    王秀娟, 陈长胜, 冯旭, 等.一阶卡尔曼滤波方法对EC集合预报气温的订正.气象灾害防御, 2019, 26(1):34-38.
    肖玉华, 赵静, 蒋丽娟.数值模式预报性能的地域性特点初步分析.暴雨灾害, 2010, 29(4):322-327.
    章大全, 郑志海, 陈丽娟, 等.10~30 d延伸期可预报性与预报方法研究进展.应用气象学报, 2019, 30(4):416-430. doi:  10.11898/1001-7313.20190403
  • 加载中
  • -->


    Figures(4)  / Tables(2)

    Article views (3462) PDF downloads(331) Cited by()
    • Received : 2020-01-15
    • Accepted : 2020-04-09
    • Published : 2020-07-31


    DownLoad:  Full-Size Img  PowerPoint