基于机器学习的高精度高分辨率气象因子时空估计

doi:10.12082/dqxxkx.2019.190014

Abstract

Abstract:

The meteorological stations are sparsely distributed across Mainland China. In terms of generating high-resolution surfaces of meteorological parameters, the estimation accuracy of existing models is limited for air temperature, and is poor for relative humidity and wind speed (few studies reported). With the measurement data of 824 monitoring stations covering the mainland of China in 2015, this study compared the typical Generalized Additive Model (GAM) and autoencoder-based residual neural network (here after, residual network for short) in terms of predicting three meteorological parameters, i.e. air temperature, relative humidity, and wind speed. The performances of the two models were evaluated through 10-fold cross-validation. Basic variables including latitude, longitude, elevation, and the day of the year are used in the air temperature models. In addition to the basic variables, the relative humidity models use air temperature and ozone concentration as covariates, and the wind speed models use wind speed coarse-resolution reanalysis data as covariates. In our spatiotemporal models, spatial coordinates capture the spatial variation and time index of the day captures the time variation. Compared with GAM, residual network significantly improved the prediction accuracy: on average, CV (Cross Validation) R²of the three meteorological factors increased by 0.21, CV RMSE decreased by 37%, and the relative humidity model improved the most (CV R²: 0.85 vs. 0.52, CV RMSE: 7.53% vs. 13.59%). With incorporation of the monthly index in the relative humidity models, the accuracy was greatly improved, indicating that the different levels of time factors are important for the relative humidity models. Furthermore, we also discussed the effectiveness and limitations of coarse resolution reanalysis data and nearest neighbor values as covariates. This study shows that the residual network model can greatly improve the accuracy of national high spatial (1 km) and temporal (daily) resolution meteorological data as opposed to traditional GAMs. Our findings provide implications for high-accuracy and high-resolution mapping of meteorological parameters in China.

Key words: meteorological factors, machine learning, residual autoencoder, Mainland China, GAM, deep learning, high resolution

Ying FANG, Lianfa LI. Spatiotemporal Estimation of High-Accuracy and High-Resolution Meteorological Parameters based on Machine Learning[J].Journal of Geo-information Science, 2019, 21(6): 799-813.DOI:10.12082/dqxxkx.2019.190014

Figures/Tables 14

Fig. 1

Tab. 1

Sources of the covariables"

数据	来源	空间分辨率	时间间隔
经度	中国地面气候资料日值数据集	-	1 d
纬度	中国地面气候资料日值数据集	-	1 d
DOY	中国地面气候资料日值数据集	-	1 d
高程	SRTM	500 m	-
风速再分析资料	GLDAS	0.25°	3 h
气温再分析资料	GLDAS	0.25°	3 h
臭氧浓度	GEOS-FP	0.25 $°$ ×0.31 $°$	3 h
相对湿度再分析资料	NCEP/NCAR	2.5°	1 d

Tab. 1

Tab. 2

Tab. 3

Tab. 4

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

Fig. 8

Fig. 9

Tab. 5

References 38

[1]	Shi Y, Ho H C, Xu Y, et al.Improving satellite aerosol optical Depth-PM2.5 correlations using land use regression with microscale geographic predictors in a high-density urban context[J]. Atmospheric Environment, 2018,190:23-34.
[2]	Gao M, Yin L, Ning J.Artificial neural network model for ozone concentration estimation and Monte Carlo analysis[J]. Atmospheric Environment, 2018,184:129-139.
[3]	Li L, Lurmann F, Habre R, et al.Constrained mixed-effect models with ensemble learning for prediction of nitrogen oxides concentrations at high spatiotemporal resolution[J]. Environmental Science & Technology, 2017,51(17):9920-9929.
[4]	王小燕,赵晓宇,陈恢富,等.江汉平原小麦孕穗期空气相对湿度升高的产量效应[J].中国农业科学,2014,47(19):3769-3779. doi: 10.3864/j.issn.0578-1752.2014.19.005
	[ Wang X Y, Zhao X Y, Chen H F, et al.Characteristics of air moisture and the effects of high air moisture at booting stage on grain yield of wheat in Jianghan plain[J]. Scientia Agricultura Sinica, 2014,47(19):3769-3779. ] doi: 10.3864/j.issn.0578-1752.2014.19.005
[5]	孙丽娜,孙瑞红,仇贵生,等.相对湿度对苹小卷叶蛾实验种群的影响[J].应用生态学报, 2014,25(12):3587-3592.
	[ Sun L N, Sun R H, Qiu G S, et al.Influence of relative humidity on the Adoxophyes orana experimental population[J]. Chinese Journal of Applied Ecology, 2014,25(12):3587-3592. ]
[6]	赵琴,潘静,曹兵,等.气温升高与干旱胁迫对宁夏枸杞光合作用的影响[J].生态学报,2015,35(18):6016-6022. doi: 10.5846/stxb201401090073
	[ Zhao Q, Pan J, Cao B, et al. Effects of elevated temperature and drought stress on photosynthesis of Lycium barbarum[J]. Acta Ecologica Sinica, 2015,35(18):6016-6022. ] doi: 10.5846/stxb201401090073
[7]	Bo Y C, Song C, Wang J, et al.Using an autologistic regression model to identify spatial risk factors and spatial risk patterns of hand, foot and mouth disease (HFMD) in Mainland China[J]. BMC Public Health, 2014,14(1):358.
[8]	谢今范,刘玉英,王玉昆,等.东北地区风能资源空间分布特征与模拟[J].地理科学,2014,34(12):1497-1503.
	[ Xie J F, Liu Y Y, Wang Y K, et al.Spatial distribution characteristics of wind resource and its simulation in northeast China[J]. Scientia Geographica Scinica, 2014,34(12):1497-1503. ]
[9]	任永建,刘敏,袁业畅,等.湖北省风能资源的高分辨率数值模拟试验[J].自然资源学报,2012,27(6):1035-1043. doi: 10.11849/zrzyxb.2012.06.013
	[ Ren Y J, Liu M, Yuan Y C, et al.The high resolution numerical simulation of wind energy resource in Hubei Province[J]. Journal of Natural Resources, 2012,26(6):1035-1043. ] doi: 10.11849/zrzyxb.2012.06.013
[10]	于洋,卫伟,陈利顶,等.黄土高原年均降水量空间插值及其方法比较[J].应用生态学报,2015,26(4):999-1006.
	[ Yu Y, Wei W, Chen L D, et al. Comparison on the methods for spatial interpolation of the annual average precipitation in the Loess Plateau region[J]. Chinese Journal of Applied Ecology, 2015,26(4):999-1006. ]
[11]	王国泰,张守平,杨清伟,等.基于空间插值方法的重庆降水信息展布[J].南水北调与水利科技,2018,16(3):18-23.
	[ Wang G T, Zhang S P, Yang Q W et al. Precipitation information distribution in Chongqing based on spatial interpolation method[J]. South-to-North Water Transfers and Water Science & Technology, 2018,16(3):18-23. ]
[12]	林忠辉,莫兴国,李宏轩,等.中国陆地区域气象要素的空间插值[J].地理学报,2002,57(1):47-56.
	[ Lin Z H, Mo X G, Li H X, et al.Comparison of three spatial interpolation methods for climate variables in China[J]. Acta Geographica Sinica, 2002,57(1):47-56. ]
[13]	张仁平,张云玲,郭靖,等.新疆地区降水分布的空间插值方法比较[J].草业科学,2018,35(3):521-529.
	[ Zhang R P, Zhang Y L, Guo J, et al.Comparison of spatial interpolation methods for precipitation distribution in Xinjiang region[J]. Pratacultural Science, 2018,35(3):521-529. ]
[14]	姜晓剑,刘小军,黄芬,等.逐日气象要素空间插值方法的比较[J].应用生态学报,2010,21(3):624-630.
	[ Jiang X J, Liu X J, Huang F, et al.Comparison of spatial interpolation methods for daily meteorological elements[J]. China Journal of Applied Ecology, 2010,21(3):624-630. ]
[15]	陈思宁,郭军.不同空间插值方法在区域气温序列中的应用评估:以东北地区为例[J].中国农业气象,2015,36(2):234-241. doi: 10.3969/j.issn.1000-6362.2015.02.015
	[ Chen S N, Guo J.Evaluation of different spatial interpolation methods in regional temperature sequence: A case study in northeast China[J]. China Journal of Agrometeorology, 2015,36(2):234-241. ] doi: 10.3969/j.issn.1000-6362.2015.02.015
[16]	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[J].地球信息科学学报,2011,13(4):539-548. doi: 10.3724/SP.J.1047.2011.00539
	[ Peng B, Zhou Y L, Gao P, et al.Suitability assessment of different interpolation methods in the gridding process of station collected air temperature: A case study in Jiangsu Province, China[J]. Journal of Geo-information Science, 2011,13(4):539-548. ] doi: 10.3724/SP.J.1047.2011.00539
[17]	李军龙,张剑,张丛,等.气象要素空间插值方法的比较分析[J].草业科学,2006(8):6-11.
	[ Li J L, Zhang J, Zhang C, et al.Analyze and compare the spatial interpolation methods for climate factor[J]. Pratacultural Science, 2006(8):6-11. ]
[18]	李伟,李庆祥,江志红.用Kriging方法对中国历史气温数据插值可行性讨论[J].南京气象学院学报,2007(2):246-252.
	[ Li W, Li Q X, Jiang Z H.Discussion on feasibility of gridding the historic temperature data in China with kriging method[J]. Journal of Nanjing Institute of Meteorology, 2007(2):246-252. ]
[19]	Xu C D, Wang J F, Li Q.A new method for temperature spatial interpolation based on sparse historical stations[J]. Journal of Climate, 2017,31(5):1757-1770.
[20]	Wang J F, Xu C, Hu M, et al.Global land surface air temperature dynamics since 1880[J]. International Journal of Climatology, 2018,381:E466-E474.
[21]	Wang J F, Xu C D, Hu M G, et al.A new estimate of the China temperature anomaly series and uncertainty assessment in 1900-2006[J]. Journal of Geophysical Research: Atmospheres, 2014,119(1):1-9.
[22]	解恒燕,张深远,侯善策,等.降水量空间插值方法在小样本区域的比较研究[J].水土保持研究,2018,25(3):117-121.
	[ Xie H Y, Zhang S Y, Hou S C, et al.Comparison research on rainfall interpolation methods for small sample areas[J]. Research of Soil and Water Conservation, 2018,25(3):1-9. ]
[23]	马秀霞,黄领梅,沈冰.陕西省月平均气温空间插值方法研究[J].水资源与水工程学报,2017,28(5):100-105.
	[ Ma X X, Huang L M, Shen B Study on spatial interpolation method of monthly mean temperature in Shanxi Province[J]. Journal of Water Resource & Water Engineering, 2017,28(5):100-105. ]
[24]	沈红,刘文兆,张勋昌.黄土高原气象要素栅格化方法的研究[J].西北农林科技大学学报(自然科学版),2010,38(4):99-106.
	[ Shen H, Liu W Z, Zhang X C.Studying the methods for rasterizing meteorological variables in the Loess Plateau[J]. Journal of Northwest A&F University(Nat. Sci. Ed.), 2010,38(4):99-106. ]
[25]	李月臣,何志明,刘春霞.基于站点观测数据的气温空间化方法评述[J].地理科学进展,2014,33(8):1019-1028. doi: 10.11820/dlkxjz.2014.08.002
	[ Li Y C, He Z M, Liu C X.Review on spatial interpolation methods of temperature data from meteorological stations[J]. Progress in Geography, 2014,33(8):1019-1028. ] doi: 10.11820/dlkxjz.2014.08.002
[26]	周婷婷,陈文惠.基于MODIS数据和气象观测数据的气温空间插值方法比较[J].地理科学进展,2011,30(9):1143-1151. doi: 10.11820/dlkxjz.2011.09.009
	[ Zhou T T, Chen W H.Comparation of the temperature spatial interpolation methods based on MODIS data and meteorological observation data[J]. Progress in Geography, 2011,30(9):1143-1151. ] doi: 10.11820/dlkxjz.2011.09.009
[27]	石志华,刘梦云,常庆瑞,等.基于优化参数的陕西省气温、降水栅格化方法分析[J].自然资源学报,2015,30(7):1141-1152.
	[ Shi Z H, Liu M Y, Chang Q R, et al.Comparison of temperature and precipitation rasterization methods based on optimized parameters in Shanxi Province[J]. Journal of Natural Resources, 2015,30(7):1141-1152. ]
[28]	姜燕敏,吴昊旻. 20个CMIP5模式对中亚地区年平均气温模拟能力评估[J].气候变化研究进展,2013,9(2):110-116. doi: 10.3969/j.issn.1673-1719.2013.02.005
	[ Jiang Y M, Wu H M.Simulation capabilities of 20 CMIP5 methods for annual mean air temperatures in central Asia[J]. Progress Inquisitions de Mutatione Climatis, 2013,9(2):110-116. ] doi: 10.3969/j.issn.1673-1719.2013.02.005
[29]	Lazzus J A.Estimation of surface soil temperature based on neural network modeling[J]. Italian Journal of Agrometeorology-rivista Italiana Di Agrometeorologia, 2014,19(2):5-12.
[30]	Szymanowski M, Kryza M, Spallek W.Regression-based air temperature spatial prediction models: An example from Poland[J]. Meteorologische Zeitschrift, 2013,22(5):577-585.
[31]	Hjort J, Suomi J, Kayhko J.Spatial prediction of urban-rural temperatures using statistical methods[J]. Theoretical and Applied Climatology, 2011,106(1-2):139-152.
[32]	Kou X, Jiang L, Bo Y, et al.Estimation of land surface temperature through blending MODIS and AMSR-E data with the Bayesian Maximum Entropy method[J]. Remote Sensing, 2016,8(2):105.
[33]	蒋冲,王飞,刘焱序,等.秦岭南北风速时空变化及突变特征分析[J].地理科学,2013,33(2):244-250.
	[ Jiang C, Wang F, Liu Y X, et al.Spatial-temporal variation and mutation of wind speed in the northern and southern regions of the Qinling Mountains[J]. Scientia Geographica Sinica, 2013,33(2):244-250. ]
[34]	赵伟,高博,刘明,等.气象因素对香港地区臭氧污染的影响[J].环境科学,2019,40(1):55-66.
	[ Zhao W, Gao B, L Q, et al. Impact of meteorological factors on the ozone pollution in Hong Kong[J]. Environmental Science, 2019,40(1):55-66. ]
[35]	Li L, Y F, J W, et al. Autoencoder based residual deep networks for robust regression prediction and spatiotemporal estimation[J].2018, preprint arXiv:1812.11262.
[36]	Wang J F, Zhang T, Fu B.A measure of spatial stratified heterogeneity[J]. Ecological Indicators, 2016,67:250-256.
[37]	王劲峰,徐成东.地理探测器:原理与展望[J].地理学报,2017,72(1):116-134.
	[ Wang J F, Xu C D.Geodector: principle and prospective[J]. Acta Geographica Sinica, 2017,72(1):116-134. ]
[38]	Wang J F, Li X H, Christakos G, et al.Geographical detectors-based health risk assessment and its application in the neural tube defects study of the Heshun region, China[J]. International Journal of Geographical Information Science, 2010,24(1):107-127.

	记录数/个	最小值	最大值	平均值	中值	标准差
气温/℃	294 357	-37.70	38.20	12.97	14.90	11.45
高程/m	294 357	1.80	4612.20	770.00	361.90	953.04
相对湿度/%	290 925	4.00	100.00	67.21	71.00	19.57
研究所得气温/℃		-18.12	38.41	12.89	13.28	10.94
GEOS-FP臭氧浓度/DU		219.40	485.40	318.30	311.80	38.40
最近邻相对湿度/）		4.00	100.00	67.28	71.00	19.61
风速/（m/s）	255 209	0.00	23.20	2.06	1.80	1.27
GLDAS风速/（m/s）	255 209	0.32	19.22	2.80	2.40	1.58

解释变量	日均气温	日均相对湿度	日均风速
经度	0.05	0.29	0.05
纬度	-0.45	-0.41	0.25
高程	-0.29	-0.35	0.13
DOY	0.13	0.20	-0.09
月份	-0.46	0.20	-0.13
研究所得气温		0.31
GEOS-FP臭氧浓度		0.15
最近邻值		0.89
GLDAS风速			0.60

	协变量组合	GAM结果				残差自编码器结果
	协变量组合	R²	RMSE	MAE		R²	RMSE	MAE
气温	经纬度+高程+DOY	0.87	4.05	3.10		0.95	2.47	1.87
	经纬度+高程+DOY+月份	0.87	4.06	3.10		0.96	2.26	1.71
相对湿度	经纬度+高程+DOY	0.51	13.77	10.96		0.72	10.37	8.05
	经纬度+高程+DOY+最近邻值	0.80	8.71	6.49		0.86		7.41	5.58
	经纬度+高程+DOY+气温	0.51	13.67	10.87		0.75		9.78	7.58
	经纬度+高程+DOY+臭氧浓度	0.52	13.59	10.79	0.75		9.74	7.55
	经纬度+高程+DOY+气温+臭氧浓度	0.52	13.64	10.81	0.77		9.47	7.29
	经纬度+高程+DOY+气温+臭氧浓度+月份	0.52	13.61	10.80	0.85		7.66	5.86
风速	经纬度+高程+DOY	0.22	11.27	7.81	0.44	9.55	6.60
风速	经纬度+高程+DOY+GEOS-FP风速	0.46	9.35	6.54	0.65		7.59	5.21
	经纬度+高程+DOY+GEOS-FP风速+月份	0.45	9.39	6.55	0.66	7.49	5.18

气象变量	模型	CV R²	CV RMSE	CV MAE
日均气温	克里格	0.92	2.87	2.14
日均气温	残差模型	0.95	2.47	1.87
日均相对湿度	克里格	0.72	8.96	6.83
日均相对湿度	残差模型	0.86	7.41	5.58
日均风速	克里格	0.14	11.78	7.67
日均风速	残差模型	0.65	7.59	5.21

Spatiotemporal Estimation of High-Accuracy and High-Resolution Meteorological Parameters based on Machine Learning

RichHTML

PDF (PC)

Like

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 14

References 38

Related Articles 15

Metrics

Comments

Recommended 0

[1]	LIU Mingjie, XU Zhuokui, GAO Yunbing, YANG Jing, PAN Yuchun, GAO Bingbo, ZHOU Yanbing, ZHOU Wanpeng, WANG Ling. Estimating Soil Organic Matter based on Machine Learning Under Sparse Sample [J]. Journal of Geo-information Science, 2020, 22(9): 1799-1813.
[2]	CHEN Ang, YANG Xiuchun, XU Bin, JIN Yunxiang, ZHANG Wenbo, GUO Jian, XING Xiaoyu, YANG Dong. Research on Recognition Methods of Elm Sparse Forest based on Object-based Image Analysis and Deep Learning [J]. Journal of Geo-information Science, 2020, 22(9): 1897-1909.
[3]	LIU Xin, ZHAO Ning, GUO Jinyun, GUO Bin. Prediction of Monthly Precipitation over the Tibetan Plateau based on LSTM Neural Network [J]. Journal of Geo-information Science, 2020, 22(8): 1617-1629.
[4]	GUO Feng, MAO Zhengyuan, ZOU Weibin, WENG Qian. A Method for Building Extraction by Fusing Feature Information from LiDAR Data and High-Resolution Imagery [J]. Journal of Geo-information Science, 2020, 22(8): 1654-1665.
[5]	LI Wan, NIU Lu, CHEN Hong, WU Hua. Robust Downscaling Method of Land Surface Temperature by Using Random Forest Algorithm [J]. Journal of Geo-information Science, 2020, 22(8): 1666-1678.
[6]	MAO Yaping, FANG Shifeng. Research of Reference Evapotranspiration's Simulation based on Machine Learning [J]. Journal of Geo-information Science, 2020, 22(8): 1692-1701.
[7]	XU Hongtao, CHEN Chunbo, ZHENG Hongwei, LUO Geping, YANG Liao, WANG Weisheng, WU Shixin. Correlation Analysis and Adaptive Genetic Algorithm based Feature Subset and Model Parameter Optimization in Salinization Monitoring [J]. Journal of Geo-information Science, 2020, 22(7): 1497-1509.
[8]	CUI Cheng, REN Hongyan, ZHAO Lu, ZHUANG Dafang. Street Space Quality Evaluation in Yuexiu District of Guangzhou City based on Multi-feature Fusion of Street View Imagery [J]. Journal of Geo-information Science, 2020, 22(6): 1330-1338.
[9]	WU Zhifeng, LUO Jiancheng, SUN Yingwei, WU Tianjun, CAO Zheng, LIU Wei, YANG Yingpin, WANG Lingyu. Research on Precision Agricultural based on the Spatial-temporal Remote Sensing Collaboration [J]. Journal of Geo-information Science, 2020, 22(4): 731-742.
[10]	GUO Yunkai, ZHANG Xiaojiong, XU Min, LIU Yuling, QIAN Jia, ZHANG Qiong. Estimation Model of Equivalent Water Thickness in the Road Area [J]. Journal of Geo-information Science, 2020, 22(2): 308-315.
[11]	LI Wenliang, WANG Chisheng, ZHU Wu. Error Spatial Distribution Characteristics of TanDEM-X 90 m DEM over China [J]. Journal of Geo-information Science, 2020, 22(12): 2277-2288.
[12]	LI Yu, ZHANG Liming, ZHANG Xingguo, WANG Hao, ZHANG Xingang. Forest Fire Risk Rapid Warning Model based on Meteorological Monitoring Network [J]. Journal of Geo-information Science, 2020, 22(12): 2317-2325.
[13]	YANG Dan, ZHOU Yanan, YANG Xianzeng, GAO Lijing, FENG Li. Vegetation Mapping in Taibai Mountain Area Supported by LSTM with Time Series Sentinel-1A Data [J]. Journal of Geo-information Science, 2020, 22(12): 2445-2455.
[14]	MAO Wenshan, ZHAO Hongli, SUN Fengjiao, JIANG Yunzhong, JIANG Qian, ZHU Yanru. Personalized Recommendation Method of Thematic Map Products based on Item2Vec with Negative Sampling Optimization [J]. Journal of Geo-information Science, 2020, 22(11): 2128-2139.
[15]	GAO Liang, DU Xin, LI Qiangzi, WANG Hongyan, ZHANG Yuan, WANG Siyuan. A Near-surface Air Temperature Spatialization Method Integrating Landuse and Soil Moisture Products [J]. Journal of Geo-information Science, 2020, 22(10): 2023-2037.