一种基于神经网络的中国区域夏季降水预测订正算法

doi:10.11755/j.issn.1006-7639(2022)-02-0308

摘要/Abstract

摘要：

基于CWRF(climate extension of WRF)区域气候模式的动力降尺度预测技术对夏季降水预测存在一定偏差,难以实现准确预测。本文立足于中国区域夏季降水特点,分析与夏季降水相关的气象要素,采用树突(dendrite,DD)网络与人工神经网络(artificial neural networks,ANN)相结合的方法,针对CWRF模式回报的1996—2019年夏季降水量进行订正,检验其订正效果。结果表明:人工树突神经网络(artificial dendritic neural network,ADNN)算法模型订正的中国夏季降水量整体好于CWRF模式历史回报,距平相关系数和时间相关系数较订正前均提高约0.10,均方误差下降约26%,趋势异常综合检验评分提高6.55,表明ADNN机器学习方法能够对CWRF模式夏季降水预测实现一定程度的订正,从而提高该模式降水预测精度。

关键词: CWRF模式, 夏季降水预测订正, DD与ANN, 均方误差, 时间相关系数, 距平相关系数

Abstract:

The prediction based on dynamic downscaling prediction technology of the climate extension of weather research and forecasting (CWRF) model to summer precipitation has a certain deviation, so it is difficult to achieve accurate prediction. This paper analyzed the correlated meteorological elements with summer precipitation based on the climatic characteristics of summer precipitation in the main land of China. And on this basis, the reforecasts of summer precipitation by CWRF model in China during 1996-2019 were corrected by using the combined method of dendritic network (DD) and artificial neural network (ANN). Finally, the correction effect was tested by mean square error (MSE), anomaly correlation coefficient (ACC) and temporal correlation coefficient (TCC), etc. The results show that the correction effect to summer precipitation based on the artificial dendritic neural network (ADNN) algorithm model was better than the historical reforecasts of CWRF model in China. The ACC and TCC both increased by about 0.10, MSE dropped by about 26%, and the overall trend anomaly test scores improved by 6.55, which indicated that the ADNN machine learning method could achieve correction to summer precipitation forecasts of CWRF model to a certain extent, thus it could improve the accuracy of precipitation forecasts of CWRF model.

Key words: CWRF model, summer precipitation forecast correction, dendritic network (DD) and artificial neural network (ANN), mean square error, temporal correlation coefficient, anomaly correlation coefficient

中图分类号:

P426.6
P456

李涛, 陈杰, 汪方, 韩锐. 一种基于神经网络的中国区域夏季降水预测订正算法[J]. 干旱气象, 2022, 40(2): 308-316.

LI Tao, CHEN Jie, WANG Fang, HAN Rui. A correction algorithm of summer precipitation prediction based on neural network in China[J]. Journal of Arid Meteorology, 2022, 40(2): 308-316.

图/表 11

图1 人工神经网络算法模型流程图

Fig.1 Flow chart of artificial neural network algorithm model

图2 人工树突神经网络算法模型流程图

Fig.2 Flow chart of artificial dendritic neural network algorithm model

图3 中国降水区域划分

Fig.3 The regional division of precipitation in China

图4 基于人工树突神经网络算法模型的夏季降水预测流程图

Fig.4 Flow chart of summer precipitation forecast based on ADNN algorithm model

图5 1996—2019年中国8个区域CWRF模式回报的夏季降水量及其ADNN算法订正值与观测值的时间相关系数 [绝对值大于等于0.33的填色区由浅至深分别通过α=0.1、0.05和0.01的显著性检验(下同)]

Fig.5 The temporal correlation coefficients of summer precipitation reforecast of CWRF model and its correction of ADNN algorithm with the observation in the eight regions of China from 1996 to 2019 (The colour areas from light to dark with absolute value greater than or equal to 0.33 passed the significance tests at 0.1, 0.05 and 0.01 level, respectively (the same as below))

图6 1996—2019年中国8个区域CWRF模式夏季回报降水量及其ADNN算法订正值与观测值的距平相关系数

Fig.6 The anomaly correlation coefficients of summer precipitation reforecast of CWRF model and its correction of ADNN algorithm with the observation in the eight regions of China from 1996 to 2019

图7 1996—2019年不同物理参数配置下CWRF模式夏季回报降水订正前后的均方误差

Fig.7 MSE of summer precipitation reforecast of CWRF model under different physical parameter configurations before and after the correction from 1996 to 2019

图8 1996—2019年中国主要大陆CWRF模式夏季回报降水量(a)及其ADNN算法订正值(b)与观测值的时间相关系数

Fig.8 The temporal correlation coefficients of summer precipitation reforecast of CWRF model (a) and its correction of ADNN algorithm (b) with the observation in main land of China from 1996 to 2019

图9 1996—2019年CWRF模式夏季回报降水量和ADNN算法订正值与观测值的距平相关系数

Fig.9 The anomaly correlation coefficients of summer precipitation reforecast of CWRF model and its correction of ADNN algorithm with the observation from 1996 to 2019

图10 1996—2019年CWRF模式夏季回报降水量和ADNN算法订正值与观测值的趋势异常综合检验评分

Fig.10 The trend abnormal comprehensive test scores of summer precipitation reforecast of CWRF model and its correction of ADNN algorithm with the observation from 1996 to 2019

图11 中国主要大陆2019年夏季降水量空间分布(单位:mm) (a)气象站观测降水,(b)CWRF模式回报降水,(c)人工树突神经网络算法模型订正降水

Fig.11 The spatial distribution of precipitation in summer of 2019 in main land of China (Unit: mm) (a) the observed precipitation at meteorological stations, (b) the reforecasted precipitation of CWRF model, (c) the corrected precipitation of ADNN algorithm model

参考文献 17

[1]	郭广芬, 杜良敏, 肖莺, 等. 长江流域夏季极端降水时空分布特征[J]. 干旱气象, 2021, 39(2):235-243.
[2]	XU H T, WU Z Y, LUO L F, et al. Verification of high-resolution medium-range precipitation forecasts from global environmental multiscale model over China during 2009-2013[J]. Atmosphere, 2018, 9(3):104. DOI URL
[3]	LIANG X Z, KUNKEL K E, SAMEL A N. Development of a regional climate model for U.S. midwest applications. Part I: sensitivity to buffer zone treatment[J]. Journal of Climate, 2001, 14(23):4363-4378. DOI URL
[4]	任绪伟, 陈晓燕, 蔡迪花, 等. GRAPES_Meso模式及其云分析系统在中国西北地区降水预报中的应用评估[J]. 干旱气象, 2021, 39(2):333-344.
[5]	刘冠州, 梁信忠. 新一代区域气候模式(CWRF)国内应用进展[J]. 地球科学进展, 2017, 32(7):781-787.
[6]	任永珍. 基于机器学习的汛期降水数据的订正算法研究[D]. 南京: 南京信息工程大学, 2020.
[7]	董晓云, 余锦华, 梁信忠, 等. CWRF模式在中国夏季极端降水模拟的误差订正[J]. 应用气象学报, 2019, 30(2):223-232.
[8]	童尧, 高学杰, 韩振宇, 等. 基于RegCM4模式的中国区域日尺度降水模拟误差订正[J]. 大气科学, 2017, 41(6):1156-1166.
[9]	张继学, 王鹏, 张琳, 等. 人工神经网络在短期降水预测方面的应用研究[J]. 科技风, 2016(17):123-124.
[10]	周雨婷, 王栋, 王远坤, 等. 典型降水预报ANN多指标优选--以太湖湖西区为例[J]. 水文, 2020, 40(1):35-39.
[11]	杜良敏, 柯宗建, 刘长征, 等. 基于聚类分区的中国夏季降水预测模型[J]. 气象, 2016, 42(1):89-96.
[12]	熊世为, 张新民, 段春锋, 等. 滁州市汛期雨量多尺度时间变化特征及其GA-BP模型预测[J]. 沙漠与绿洲气象, 2019, 13(5):89-93.
[13]	LI X, TANG J, ZHANG Q, et al. Power-efficient neural network with artificial dendrites[J]. Nature Nanotechnology, 2020, 15(9):776-782. DOI URL
[14]	夏克文, 李昌彪, 沈钧毅. 前向神经网络隐含层节点数的一种优化算法[J]. 计算机科学, 2005, 32(10):143-145.
[15]	LIU G, WANG J. Dendrite net: a white-box module for classification, regression, and system identification[J]. IEEE Transactions on Cybernetics, 2021. Doi: 10.1109/TCYB.2021.3124328. DOI
[16]	李宁, 曾新民, 席朝笠. 我国华东地区月尺度气候动力预测的研究:(Ⅰ)区域模式建立及模拟检验[J]. 气象科学, 2006, 26(6):591-597.
[17]	WANG F, YANG S. Regional characteristics of long-term changes in total and extreme precipitations over China and their links to atmospheric-oceanic features[J]. International Journal of Climatology, 2017, 37(2):751-769. DOI URL