基于机器学习的参考作物蒸散量估算研究

doi:10.12082/dqxxkx.2020.200085

地球信息科学学报 ›› 2020, Vol. 22 ›› Issue (8): 1692-1701.doi: 10.12082/dqxxkx.2020.200085

基于机器学习的参考作物蒸散量估算研究

毛亚萍¹^,²(), 房世峰¹^,^*()

1.中国科学院地理科学与资源研究所资源与环境信息系统国家重点实验室,北京 100101
2.中国科学院大学,北京 100049

收稿日期:2020-02-18 修回日期:2020-04-16 出版日期:2020-08-25 发布日期:2020-10-25
作者简介:毛亚萍（1992— ）,女,四川新津人,硕士生,主要从事环境物联网及其应用研究。E-mail: maoyp.17s@igsnrr.ac.cn
基金资助:
中国科学院战略性先导科技专项(XDA20010302);国家自然科学基金项目(41971082);国家自然科学基金项目(U1503184)

Research of Reference Evapotranspiration's Simulation based on Machine Learning

MAO Yaping¹^,²(), FANG Shifeng¹^,^*()

1. State Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China
2. University of Chinese Academy of Sciences, Beijing 100049, China

Received:2020-02-18 Revised:2020-04-16 Online:2020-08-25 Published:2020-10-25
Contact: FANG Shifeng
Supported by:
Strategic Priority Research Program of the Chinese Academy of Sciences(XDA20010302);National Natural Science Foundation of China(41971082);National Natural Science Foundation of China(U1503184)

摘要/Abstract

摘要：

参考作物蒸散量（Reference Evapotranspiration, ET₀）的准确估算对区域水资源管理和分配、流域水量平衡以及气候变化等研究具有重要作用。新疆地处我国西北干旱地区,水资源供需矛盾尖锐,精确估算该地区的ET₀有助于其科学合理地调配水资源,缓解水资源供需压力。FAO推荐的Penman-Monteith法是计算ET₀的标准方法,但该方法需要多项气象因子,而新疆地区气象站点较少且分布不均,精确完备的气象数据在新疆大部分区域难以获取。因此,如何使用有限气象因子获取高精度的ET₀在新疆地区备受关注。本文基于中国气象数据网提供的新疆地区1980—2019年的地面气候资料日值数据集,在日和月尺度下,通过对最高气温T_max、最低气温T_min、平均气温T_avg、风速U₂、相对湿度RH和日照时数n共6项气象因子进行敏感性分析,形成不同的气象因子组合;然后使用SVM、RF、GBDT和ELM 4种机器学习算法,以FAO-56 PM计算值为标准值,对新疆地区的ET₀进行了拟合建模;最后,从拟合精度、稳定性和计算代价3个方面对模型进行评价。研究表明：① 在新疆地区,ET₀对RH、T_max和U₂敏感系数级别为高,平均敏感系数分别为-0.516、0.283和0.266;n为中等,平均敏感系数为0.124;T_min和T_avg为低,平均敏感系数分别为-0.016和-0.003;② 在日尺度,各算法在RH、T_max、U₂和n这4项气象因子为输入时精度较高（RMSE<0.5 mm/day,R²>0.95）,可对ET₀进行精确估算;在月尺度,各算法使用RH、T_max和U₂这3项参数便可对ET₀进行精确估算。SVM和GBDT这2种算法在日尺度和月尺度都有较好的适用性,可在相应尺度下使用较少气象因子替代FAO-56 PM公式对ET₀进行估算。

关键词: 参考作物蒸散量, 机器学习, Penman-Monteith, 新疆, 支持向量机, 随机森林, 梯度提升树, 极限学习机

Abstract:

Accurate estimation of Reference Evapotranspiration (ET₀) is essential to agricultural water management and allocation and hydrological cycle research. FAO-56 Penman-Monteith (FAO-56 PM) is the standard method to calculate ET₀ recommended by Food and Agriculture Organization of the United Nations (FAO). But this method demands too many parameters and these meteorological inputs are not commonly available or unreliable, especially in Xinjiang province. Under this situation, machine learning algorithms have been introduced to estimate ET₀ using fewer meteorological parameters and many comparisons of their prediction accuracy have been conducted. But the input combinations of meteorological factors are various and lack theoretical support. Meanwhile, the comparison of their performance at different time-scales has not been comprehensively conducted yet, and the good stability and less computational effort of models are also less to consider. The objective of this research was to evaluate machine learning algorithms' performance in modeling daily ET₀and monthly ET₀using fewer meteorological factors in Xinjiang. At this point, by using data collected from 41 weather stations in Xinjiang, this paper used Sensitivity Coefficient (S_V) to evaluate the meteorological factors' influence degree to ET₀ and then combined factors with high influence as input to Support Vector Machine (SVM), Gradient Boosted Decision Tree (GBDT), Random Forest (RF), and Extreme Learning Machine (ELM) in modeling daily and monthly ET₀, and finally investigated and compared the performance of these algorithms from accuracy, stability and computational cost. The results showed RH (S_V=-0.516), T_max(S_V=0.283)and U₂ (S_V=0.266) had high influence to ET₀followed by n (S_V=0.124), while T_min (S_V= -0.016) and T_avg (S_V= -0.003) exhibited low influence. In modeling daily ET₀, models obtained satisfactory accuracy (RMSE<0.5 mm/day, R²>0.95) with input combination of RH, T_max, U₂ and n, while combination of RH、T_max and U₂ showed comparable accuracy for monthly ET₀prediction. The SVM and GBDT models showed the best performance and have been recommended for daily and monthly ET₀estimation in Xinjiang and maybe elsewhere with similar climates around the world.

Key words: Reference Evapotranspiration, machine learning, Penman-Monteith, Xinjiang, Support Vector Machine, Random Forest, Gradient Boosted Decision Tree, Extreme Learning Machine

毛亚萍, 房世峰. 基于机器学习的参考作物蒸散量估算研究[J]. 地球信息科学学报, 2020, 22(8): 1692-1701.DOI:10.12082/dqxxkx.2020.200085

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.DOI:10.12082/dqxxkx.2020.200085

图/表 10

图1

表1

图2

表2

表3

表4

表5

表6

图3

图4

参考文献 25

[1]	Allen R G, Pereira L S, Raes D, et al. Crop evapotranspiration- guidelines for computing crop water requirements[M]. Rome: Food and Agriculture Organization of the United Nations, 1998.
[2]	Wen X, Si J, He Z, et al. Support-vector-machine-based models for modeling daily reference evapotranspiration with limited climatic data in extreme arid regions[J]. Water Resources Management, 2015,29(9):3195-3209. doi: 10.1007/s11269-015-0990-2
[3]	Wu L, Fan J. Comparison of neuron-based, kernel-based, tree-based and curve-based machine learning models for predicting daily reference evapotranspiration[J]. PLoS One, 2019,14(5):e0217520. doi: 10.1371/journal.pone.0217520 pmid: 31150448
[4]	王升, 付智勇, 陈洪松, 等. 基于随机森林算法的参考作物蒸发蒸腾量模拟计算[J]. 农业机械学报, 2017,48(3):302-309.
	[ Wang S, Fu Z Y, Chen H S, et al. Simulation of reference evapotranspiration based on random forest method[J]. Transactions of the Chinese Society for Agricultural Machinery, 2017,48(3):302-309. ]
[5]	Huang G, Wu L, Ma X, et al. Evaluation of CatBoost method for prediction of reference evapotranspiration in humid regions[J]. Journal of Hydrology, 2019,574:1029-1041. doi: 10.1016/j.jhydrol.2019.04.085
[6]	Fan J, Ma X, Wu L, et al. Light gradient boosting machine: an efficient soft computing model for estimating daily reference evapotranspiration with local and external meteorological data[J]. Agricultural Water Management, 2019,225:105758. doi: 10.1016/j.agwat.2019.105758
[7]	Fan J, Yue W, Wu L, et al. Evaluation of SVM, ELM and four tree-based ensemble models for predicting daily reference evapotranspiration using limited meteorological data in different climates of China[J]. Agricultural and Forest Meteorology, 2018,263:225-241. doi: 10.1016/j.agrformet.2018.08.019
[8]	赵文刚, 马孝义, 刘晓群, 等. 基于神经网络算法的广东省典型代表站点ET₀简化计算模型研究[J]. 灌溉排水学报, 2019,38(5):91-99.
	[ Zhao W G, Ma X Y, Liu X Q, et al. Using neural network model to simplify ET₀ calculation for representative stations in Guangdong province[J]. Journal of Irrigation and Drainage, 2019,38(5):91-99. ]
[9]	Feng Y, Peng Y, Cui N, et al. Modeling reference evapotranspiration using extreme learning machine and generalized regression neural network only with temperature data[J]. Computers and Electronics in Agriculture, 2017,136:71-78. doi: 10.1016/j.compag.2017.01.027
[10]	Ferreira L B, Da Cunha F F, De Oliveira R A, et al. Estimation of reference evapotranspiration in Brazil with limited meteorological data using ANN and SVM: A new approach[J]. Journal of Hydrology, 2019,572:556-570. doi: 10.1016/j.jhydrol.2019.03.028
[11]	张皓杰, 崔宁博, 徐颖, 等. 基于ELM的西北旱区参考作物蒸散量预报模型[J]. 排灌机械工程学报, 2018,36(8):779-784.
	[ Zhang H J, Cui N B, Xu Y, et al. Prediction for reference crop evapotranspiration in arid northwest China based on ELM[J]. Journal of Drainage and Irrigation Machinery Engineering, 2018,36(8):779-784. ]
[12]	吴立峰, 鲁向晖, 刘小强, 等. 蝙蝠算法优化极限学习机模拟参考作物蒸散量[J]. 排灌机械工程学报, 2018,36(9):802-805,829.
	[ Wu L F, Lu X H, Liu X Q, et al. Simulation of reference crop evapotranspiration by using bat algorithm optimization based extreme learning machine[J]. Journal of Drainage and Irrigation Machinery Engineering, 2018,36(9):802-805,829. ]
[13]	冯禹, 崔宁博, 龚道枝, 等. 基于极限学习机的参考作物蒸散量预测模型[J]. 农业工程学报, 2015,31(S1):153-160.
	[ Feng Y, Cui N, Gong D, et al. Prediction model of reference crop evapotranspiration based on extreme learning machine[J]. Transactions of the Chinese Society of Agricultural Engineering, 2015,31(S1):153-160. ]
[14]	Wu L, Zhou H, Ma X, et al. Daily reference evapotranspiration prediction based on hybridized extreme learning machine model with bio-inspired optimization algorithms: application in contrasting climates of China[J]. Journal of Hydrology, 2019,577:123960. doi: 10.1016/j.jhydrol.2019.123960
[15]	Reis M M, Da Silva A J, Zullo Junior J, et al. Empirical and learning machine approaches to estimating reference evapotranspiration based on temperature data[J]. Computers and Electronics in Agriculture, 2019,165:104937. doi: 10.1016/j.compag.2019.104937
[16]	刘悦, 崔宁博, 李果, 等. 近56年西南地区四季参考作物蒸散量变化成因分析[J]. 节水灌溉, 2018(12):54-59.
	[ Liu Y, Cui N B, Li G, et al. Attribution analysis of seasonal reference crop evapotranspiration in southwest China in recent 56 years[J]. Water Saving Irrigation, 2018(12):54-59. ]
[17]	Lenhart T, Eckhardt K, Fohrer N, et al. Comparison of two different approaches of sensitivity analysis[J]. Physics and Chemistry of the Earth, 2002,27(9-10):645-654.
[18]	周志华. 机器学习[M]. 北京: 清华大学出版社, 2016: 133-139.
	[ Zhou Z H. Machine learning[M]. Beijing: Tsinghua University Press, 2016: 133-139. ]
[19]	Kisi O. Pan evaporation modeling using least square support vector machine, multivariate adaptive regression splines and M5 model tree[J]. Journal of Hydrology, 2015,528:312-320. doi: 10.1016/j.jhydrol.2015.06.052
[20]	Breiman L. Random forests[J]. Machine Learning, 2001,45(1):5-32. doi: 10.1023/A:1010933404324
[21]	李航. 统计学习方法[M]. 北京: 清华大学出版社, 2012: 55-75.
	[ Li H. Statistical learning method[M]. Beijing: Tsinghua University Press, 2012: 55-75. ]
[22]	张雷, 王琳琳, 张旭东, 等. 随机森林算法基本思想及其在生态学中的应用——以云南松分布模拟为例[J]. 生态学报, 2014,34(3):650-659. doi: 10.5846/stxb201306031292
	[ Zhang L, Wang L L, Zhang X D, et al. The basic principle of random forest and its applications in ecology: A case study of Pinus yunnanensis[J]. Acta Ecologica Sinica, 2014,34(3):650-659. ]
[23]	Friedman J H. Stochastic gradient boosting[J]. Computational Statistics & Data Analysis, 2002,38(4):367-378.
[24]	Huang G B, Zhu Q Y, Siew C K, et al. Extreme learning machine: a new learning scheme of feedforward neural networks. Proceedings of the 2004 Ieee International Joint Conference on Neural Networks[C]. New York:IEEE, 2004: 985-990.
[25]	Hassan M A, Khalil A, Kaseb S, et al. Exploring the potential of tree-based ensemble methods in solar radiation modeling[J]. Applied Energy, 2017,203:897-916. doi: 10.1016/j.apenergy.2017.06.104

尺度	数据量 /条	平均值(1980-01-01—2019-12-31)
尺度	数据量 /条	T_avg/℃	T_max/℃	T_min/℃	U₂/m·s^-1	RH/%	n/h
日	19 742	8.09	14.98	2.08	1.68	51.49	7.76
月	19 680	8.53	15.42	2.49	1.69	51.16	7.83

参数类别	具体参数
日期	年内日序数J
地理位置参数	高程z/m、纬度φ
气象参数	日最高气温T_max/℃、日最低气温T_min/℃、日平均气温T_avg/℃、日平均风速U₂/(m/s)、日平均相对湿度RH/%、日照时数n/h

组合	输入参数
Group 1	RH、T_max
Group 2	RH、U₂
Group 3	RH、T_max、U₂
Group 4	RH、T_max、U₂、n

算法	气象组合	RMSE/(mm·day^-1)		R²
算法	气象组合	日尺度	月尺度	日尺度	月尺度
SVM	Group1	0.934	0.693	0.840	0.899
	Group2	1.655	1.418	0.500	0.575
	Group3	0.521	0.264	0.950	0.985
	Group4	0.392	0.239	0.972	0.988
RF	Group1	1.076	0.736	0.788	0.885
	Group2	1.773	1.500	0.426	0.524
	Group3	0.563	0.270	0.942	0.985
	Group4	0.412	0.241	0.969	0.988
GBDT	Group1	0.939	0.680	0.839	0.902
	Group2	1.624	1.380	0.518	0.597
	Group3	0.541	0.275	0.947	0.984
	Group4	0.425	0.257	0.967	0.986
ELM	Group1	0.947	0.710	0.836	0.893
	Group2	1.625	1.411	0.517	0.579
	Group3	0.544	0.290	0.946	0.982
	Group4	0.434	0.274	0.966	0.984

基于机器学习的参考作物蒸散量估算研究

Research of Reference Evapotranspiration's Simulation based on Machine Learning

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 25

相关文章 15

Metrics

本文评价

推荐阅读 0

尺度	T_avg	T_max	T_min	U₂	RH	n
日	-0.001	0.28	-0.011	0.266	-0.529	0.107
月	-0.005	0.286	-0.021	0.267	-0.503	0.141

[1]	戚梦, 陈楠, 林偲蔚, 周千千. 引入集水区复杂网络的中国地貌识别研究[J]. 地球信息科学学报, 2023, 25(5): 909-923.
[2]	李琳叶, 李艳艳, 陈传法, 刘妍, 刘雅婷, 刘盼盼. 林区数字高程模型修正方法：顾及高程自相关的后向传播神经网络模型[J]. 地球信息科学学报, 2023, 25(5): 935-952.
[3]	林炫歆, 肖桂荣, 周侯伯. 顾及土地利用动态变化的滑坡易发性评估方法[J]. 地球信息科学学报, 2023, 25(5): 953-966.
[4]	柯日宏, 吴升, 柯玮文. 一种识别共享单车潮汐点的时空模型和基于KNN-LightGBM的租还需求预测方法[J]. 地球信息科学学报, 2023, 25(4): 741-753.
[5]	李心雨, 闫浩文, 王卓, 王炳瑄. 街景图像与机器学习相结合的道路环境安全感知评价与影响因素分析[J]. 地球信息科学学报, 2023, 25(4): 852-865.
[6]	秦肖伟, 程博, 杨志平, 李林, 董文, 张新, 杨树文, 靳宗义, 薛庆. 基于时序遥感影像的西南山区地块尺度作物类型识别[J]. 地球信息科学学报, 2023, 25(3): 654-668.
[7]	黄帅元, 董有福, 李海鹏. 黄土高原区SRTM1 DEM高程误差校正模型构建及对比分析[J]. 地球信息科学学报, 2023, 25(3): 669-681.
[8]	潘雨飘, 赵翔, 王静, 张亦清, 刘耀林. 基于SMOTE-RF算法的村庄发展类型识别方法研究[J]. 地球信息科学学报, 2023, 25(1): 163-176.
[9]	杨洋, 邵哲平, 赵强, 潘家财, 胡雨, 梅强. 基于厦门港的海上交通事故地理空间分布及风险预测研究[J]. 地球信息科学学报, 2022, 24(9): 1676-1687.
[10]	吴亚楠, 郭长恩, 于东平, 段爱民, 刘玉, 董士伟, 单东方, 吴耐明, 李西灿. 基于不确定性分析的遥感分类空间分层及评估方法[J]. 地球信息科学学报, 2022, 24(9): 1803-1816.
[11]	聂祥琴, 陈瀚阅, 牛铮, 张黎明, 刘炜, 邢世和, 范协裕, 李家国. 基于时序影像的农业活动因子提取与闽西耕地SOC数字制图[J]. 地球信息科学学报, 2022, 24(9): 1835-1852.
[12]	高位, 涂伟, 李明晓, 方碧宸, 陈栋胜, 黄正东, 贺彪. 基于影响因子动态更新的都市圈国土-人口协同模拟[J]. 地球信息科学学报, 2022, 24(8): 1502-1511.
[13]	裴泽华, 葛淼, 李浩, 何进伟, 王聪霞. 基于随机森林模型的中国中老年人群HDL-C环境影响因素研究[J]. 地球信息科学学报, 2022, 24(7): 1286-1300.
[14]	郭逸飞, 吴田军, 骆剑承, 石含宁, 郜丽静. 基于不确定性迭代优化的山地植被遥感制图[J]. 地球信息科学学报, 2022, 24(7): 1406-1419.
[15]	高雪梅, 杨续超, 陈柏儒, 林琳. 基于随机森林模型的环渤海地区人口空间化模拟[J]. 地球信息科学学报, 2022, 24(6): 1150-1162.