Crop yield prediction in Ethiopia based on machine learning under future climate scenarios
-
摘要: 对于以农业产业为支柱的埃塞俄比亚, 粮食供应和安全对国家安全和人民的生计尤为重要。由于作物生长和气候因素之间的复杂耦合关系, 预测气候变化对农作物产量影响具有较大难度, 机器学习技术则为预测这种复杂系统的变化提供了一种有效途径。本研究利用37个全球气候模式(GCM)的数据以及土壤数据, 基于机器学习模型, 预测了埃塞俄比亚2021年至2050年5种粮食作物在SSP126、SSP245和SSP585情景下的产量变化。在经过GCM和变量的筛选后, 利用梅赫季和贝尔格季中5种作物共10个产量数据对直方图梯度提升决策树、极端梯度提升随机森林、轻梯度提升决策树、随机森林、极限树以及K近邻6种机器学习模型进行训练。经过模型评估, 选择表现良好的3个模型并采用线性回归算法进行堆叠, 然后使用模型进行预测。研究结果表明, 埃塞俄比亚在未来30年间, 梅赫季各个作物的产量变化都以增产少于2 t·hm−2为主, 而在SSP126情景下的贝尔格季将出现更明显的产量减少, 这可能是由于温室效应的减缓降低了CO2的施肥效应。随着社会矛盾加剧和人类活动造成的生态环境恶化, 研究区的农业对改变粮食农业结构和重新分配生产力的需求不断增长, 导致农作物生产力向新的适宜地区转移。在SSP126和SSP585情景下, 研究区域将分别因为干旱情况的和温室效应加剧而获得更高的粮食作物生产力。Abstract: Crop yield and agricultural development are the foundation of human survival. In Ethiopia, where agriculture is the economic backbone, food supply and security are crucial for national security and people's livelihoods. Crop yield is greatly influenced by climatic conditions, but the coupling relationship between them has not been clearly explained, which poses difficulties for quantitatively analyzing crop yields under climate change. The development of machine learning techniques provides a method for predicting changes in such complex systems. This study predicts the changes in the yield of five major staple crops in Ethiopia from 2021 to 2050 by using machine learning methods combined with climate predictions from Global Climate Models (GCMs) under different future scenarios in the Sixth Coupled Model Intercomparison Project (CMIP6). First, data on 9 climate variables from 37 GCMs under four scenarios (historical, SSP1-2.6, SSP2-4.5 and SSP5-5.8) in CMIP6 were obtained. A Taylor diagram was used to select the best-performing GCMs and calculate their weighted averages. These averages were combined with five soil indicators to form an independent variable database. After removing highly correlated variables using Spearman's correlation coefficient, machine learning models were trained using 10 yield data variables of teff, maize, wheat, barley and sorghum for two major growing seasons in Ethiopia from 1995 to 2020 as dependent variables. This paper employed Histogram Gradient Boosting (HGB), Extreme Gradient Boosting Random Forest (XGBRF), Light Gradient Boosting Machine (LGBM), Random Forest (RF), Extra Trees (ET) and K-Neighbors as machine learning models. After model evaluation, the top-performing three models were stacked using linear regression. The independent variables were input into the final model to predict the yields of the 10 staple crops in Ethiopia from 2021 to 2050. The results were analyzed, and the following conclusions were drawn. (1) CMCC-CM2-SR5, MPI-ESM1-2-LR, EC-Earth3-Veg-LR, EC-Earth3-Veg and MPI-ESM1-2-HR obtained higher overall scores in the Taylor diagram analysis, indicating better simulation of climate in Ethiopian compared to other GCMs. (2) The Correlation coefficient (R2), Mean absolute error (MAE), and Explained variance score (EVS) of the XGBRF, RF and ET were higher than those of HGB, LGBM and K-neighbors. The stacking method of ensemble learning improved the performance of the ensemble model over individual models. (3) Over the next 30 years, the changes in crop yield during the Meher season (the longer growing season in Ethiopia, which is generally from April to December) were mainly within 2 t·hm−2. In the Belg season (the shorter growing season in Ethiopia, which is generally from February to September), there was a greater decrease in yield under SSP126 scenario, while the other two scenarios showed an increase, possibly due to the mitigation of greenhouse effects reducing the fertilization effect of CO2. (4) With intensification of social conflicts and environmental degradation caused by human activities, there is a growing need in the research area to change the agricultural structure and redistribute productivity, and this leads to the transfer of agricultural productivity to new suitable areas. Under SSP126 and SSP585 scenarios, the research area will achieve higher crop productivity due to the alleviation of drought conditions and the exacerbation of greenhouse effects, respectively. The results of this study demonstrate the changes in crop yield in the research area under different future climate change scenarios, providing references for determining agricultural production potential and formulating agricultural policies in the research area.
-
Key words:
- Crop yield /
- Machine learning /
- Climate change /
- Global Climate Model /
- Ethiopia
-
图 2 37个GCM近地面气压变量(a)和近地面气温变量(b)的泰勒图
ps 和tas分别表示近地面气压和近地面气温。ps and tas represent surface air pressure and near-surface air temperature, respectively. AE5: ACCESS-ESM1-5; AC11M: AWI-CM-1-1-MRAE11L: AWI-ESM-1-1-LR; BCM: BCC-CSM2-MR; BE: BCC-ESM1; CC0: CAMS-CSM1-0; CE: CanESM5; CE0: CAS-ESM2-0; CWF: CESM2-WACCM-FV2; C: CIESM; CCH: CMCC-CM2-HR4; CCS: CMCC-CM2-SR5; E11E: E3SM-1-1-ECA; EE: EC-Earth3; EEV: EC-Earth3-Veg; EEVL: EC-Earth3-Veg-LR; FE20: FIO-ESM-2-0; GE: GFDL-ESM4; GE1G: GISS-E2-1-G; GE1H: GISS-E2-1-H: ; IE: IITM-ESM; IC8: INM-CM4-8; IC0: INM-CM5-0; ICL: IPSL-CM6A-LR; K10G: KACE-1-0-G; M: MIROC6; ME12H: MPI-ESM-1-2-HAM; ME2H: MPI-ESM1-2-HR; ME2L: MPI-ESM1-2-LR; ME0: MRI-ESM2-0; N: NESM3; NC: NorCPM1; NL: NorESM2-LM; NM: NorESM2-MM; S U: SAM0-UNICON; TE: TaiESM1)
Figure 2. Taylor diagram of surface air pressure (a) and near-surface air temperature (b) for 37 GCM
图 3 14个变量和5种梅赫季作物产量的斯皮尔曼相关性热图
huss: 近地面比湿度 Near-Surface specific humidity; pr: 降水量 Precipitation; ps: 近地面气压 Surface air pressure; rsds: 地面下行短波辐射量 Surface downwelling shortwave radiation; tas: 近地面气温 Near-Surface air temperature; tasmax: 近地面日最高气温 Daily maximum near-Surface air temperature; tasmin: 近地面日最低气温 Daily minimum near-Surface air temperature; uas: 近地面东向风速 Eastward near-Surface wind; vas: 近地面北向风速 Northward near-Surface wind; tn: 总氮 Total nitrogen; bd: 容重 Bulk density; cec: 阳离子交换量 Cation exchange capacity; soc: 土壤有机碳 Soil organic content; ph: pH; b_m: 大麦(梅赫季) Barley (Meher Season); w_m: 小麦(梅赫季) Wheat (Meher Season); m_m: 玉米(梅赫季) Maize (Meher Season); s_m: 高粱(梅赫季) Sorghum (Meher Season); t_m: 苔麸(梅赫季) Teff (Meher Season)。
Figure 3. Spearman correlation heat map with 14 variables and yield of 5 crops in Meher season
表 1 研究初步选用的未来气候变化模式气候参数列表及信息
Table 1. List and information of preliminarily selected GCM variables
名称 Variable name 代号 Alias 单位Unit 近地面比湿度 Near-Surface specific humidity huss 1 降水量 Precipitation pr kg∙m−2∙s−1 近地面气压 Surface air pressure ps Pa 地面下行短波辐射量 Surface downwelling shortwave radiation rsds W∙m−2 近地面气温 Near-Surface air temperature tas K 近地面日最高气温 Daily maximum near-Surface air temperature tasmax K 近地面日最低气温 Daily minimum near-Surface air temperature tasmin K 近地面东向风速 Eastward near-Surface wind uas m∙s−1 近地面北向风速 Northward near-Surface wind vas m∙s−1 表 2 6种模型的模拟表现
Table 2. Prediction performance of 6 different models
模型名称
Model name判决系数
Coefficient of Determination, R2平均绝对误差
Mean Absolute Error, MAE解释方差评分
Explained Variance Score, EVS直方图梯度提升
Histogram Gradient Boosting (HGB)0.51 2.33 0.38 极端梯度提升随机森林
Extreme Gradient Boosting Random Forest (XGBRF)0.74 3.02 0.69 轻梯度提升
Light Gradient Boosting Machine (LGBM)0.64 8.24 0.68 随机森林 Random Forest (RF) 0.69 3.64 0.65 极限树 Extra Trees (ET) 0.67 1.65 0.59 K近邻 K-Neighbors 0.55 5.65 0.42 表 3 最终模型对不同作物的模拟表现
Table 3. Prediction performance of the stacked model for different crops
作物种类
Crop type判决系数
(Coefficient of Determination, R2)平均绝对误差
(Mean Absolute Error, MAE)解释方差评分
(Explained Variance Score, EVS)苔麸(梅赫季)
Teff (Meher Season)0.76 2.35 0.64 玉米(梅赫季)
Maize (Meher Season)0.88 1.22 0.91 小麦(梅赫季)
Wheat (Meher Season)0.82 2.46 0.71 大麦(梅赫季)
Barley (Meher Season)0.73 3.15 0.72 高粱(梅赫季)
Sorghum (Meher Season)0.71 4.3 0.75 苔麸(贝尔格季)
Teff (Belg Season)0.60 4.57 0.55 玉米(贝尔格季)
Maize (Belg Season)0.67 4.69 0.56 小麦(贝尔格季)
Wheat (Belg Season)0.62 4.25 0.60 大麦(贝尔格季)
Barley (Belg Season)0.31 19.53 0.12 高粱(贝尔格季)
Sorghun (Belg Season)0.59 5.80 0.58 -
[1] PATEL R. The long green revolution[J]. Journal of Peasant Studies, 2013, 40(1): 1−63 doi: 10.1080/03066150.2012.719224 [2] TIRADO M C, CRAHAY P, MAHY L, et al. Climate change and nutrition: creating a climate for nutrition security[J]. Food and Nutrition Bulletin, 2013, 34(4): 533−547 doi: 10.1177/156482651303400415 [3] CONNOLLY-BOUTIN L, SMIT B. Climate change, food security, and livelihoods in sub-Saharan Africa[J]. Regional Environmental Change, 2016, 16(2): 385−399 doi: 10.1007/s10113-015-0761-x [4] KENNEDY E, JAFARI A, STAMOULIS K G, et al. The first Programmefood and nutrition security, impact, resilience, sustainability and transformation: review and future directions[J]. Global Food Security, 2020, 26: 100422 doi: 10.1016/j.gfs.2020.100422 [5] WUDIL A H, USMAN M, ROSAK-SZYROCKA J, et al. Reversing years for global food security: a review of the food security situation in sub-saharan Africa (SSA)[J]. International Journal of Environmental Research and Public Health, 2022, 19(22): 14836 doi: 10.3390/ijerph192214836 [6] ZHANG W Q, PERSOZ L, HAKIZA S, et al. Impact of COVID-19 on food security in Ethiopia[J]. Epidemiologia (Basel, Switzerland), 2022, 3(2): 161−178 [7] BROWN M E, FUNK C C. Climate. Food security under climate change[J]. Science, 2008, 319(5863): 580−581 doi: 10.1126/science.1154102 [8] GREGORY P J, INGRAM J S, BRKLACICH M. Climate change and food security[J]. Philosophical Transactions of the Royal Society B:Biological Sciences, 2005, 360(1463): 2139−48 doi: 10.1098/rstb.2005.1745 [9] PARRY M, ROSENZWEIG C, IGLESIAS A, et al. Climate change and world food security: a new assessment[J]. Global Environmental Change, 1999, 9: S51−S67 doi: 10.1016/S0959-3780(99)00018-7 [10] WHEELER T, VON BRAUN J. Climate change impacts on global food security[J]. Science, 2013, 341(6145): 508−513 doi: 10.1126/science.1239402 [11] 张丽娟, 姚子艳, 唐世浩, 等. 20世纪80年代以来全球耕地变化的基本特征及空间格局[J]. 地理学报, 2017, 72(7): 1235−1247ZHANG L J, YAO Z Y, TANG S H, et al. Spatiotemporal characteristics and patterns of the global cultivated land since the 1980s[J]. Acta Geographica Sinica, 2017, 72(7): 1235−1247 [12] AGGARWAL P K, BANERJEE B, DARYAEI M G, et al. InfoCrop: a dynamic simulation model for the assessment of crop yields, losses due to pests, and environmental impact of agro-ecosystems in tropical environments. II. Performance of the model[J]. Agricultural Systems, 2006, 89(1): 47−67 doi: 10.1016/j.agsy.2005.08.003 [13] AGGARWAL P K, KALRA N, CHANDER S, et al. InfoCrop: a dynamic simulation model for the assessment of crop yields, losses due to pests, and environmental impact of agro-ecosystems in tropical environments. I. Model description[J]. Agricultural Systems, 2006, 89(1): 1−25 doi: 10.1016/j.agsy.2005.08.001 [14] CHALLINOR A J, WHEELER T R. Crop yield reduction in the tropics under climate change: processes and uncertainties[J]. Agricultural and Forest Meteorology, 2008, 148(3): 343−356 doi: 10.1016/j.agrformet.2007.09.015 [15] DHUNGANA P, ESKRIDGE K M, WEISS A, et al. Designing crop technology for a future climate: an example using response surface methodology and the CERES-Wheat model[J]. Agricultural Systems, 2006, 87(1): 63−79 doi: 10.1016/j.agsy.2004.11.004 [16] GBETIBOUO G A, HASSAN R M. Measuring the economic impact of climate change on major South African field crops: a Ricardian approach[J]. Global and Planetary Change, 2005, 47(2/3/4): 143−152 [17] HOOGENBOOM G. Contribution of agrometeorology to the simulation of crop production and its applications[J]. Agricultural and Forest Meteorology, 2000, 103(1/2): 137−157 [18] HOWDEN S M, O'LEARY G J. Evaluating options to reduce greenhouse gas emissions from an Australian temperate wheat cropping system[J]. Environmental Modelling & Software, 1997, 12(2/3): 169−176 [19] 杨文龙, 杜德斌, 刘承良, 等. 中国地缘经济联系的时空演化特征及其内部机制[J]. 地理学报, 2016, 71(6): 956−969YANG W L, DU D B, LIU C L, et al. Temporal and spatial evolution characteristics and internal mechanism of geo-economic relations in China[J]. Acta Geographica Sinica, 2016, 71(6): 956−969 [20] 唐国平, 李秀彬, FISCHER G, et al. 气候变化对中国农业生产的影响[J]. 地理学报, 2000(02): 129−138 [21] 赵茹欣, 王会肖, 董宇轩. 气候变化对关中地区粮食产量的影响及趋势分析[J]. 中国生态农业学报(中英文), 2020, 28(4): 467−479ZHAO R X, WANG H X, DONG Y X. Impact of climate change on grain yield and its trend across Guanzhong region[J]. Chinese Journal of Eco-Agriculture, 2020, 28(4): 467−479 [22] HERRERA-PANTOJA M, HISCOCK K M. The effects of climate change on potential groundwater recharge in Great Britain[J]. Hydrological Processes, 2008, 22(1): 73−86 doi: 10.1002/hyp.6620 [23] 倪宁淇, 谢佳鑫, 刘小莽, 等. 基于径流对气候变化敏感性指标的多源数据质量评估[J]. 地理学报, 2022, 77(9): 2280−2291NI N Q, XIE J X, LIU X M, et al. Multi-source data quality assessment based on the index of runoff sensitivity to climate change[J]. Acta Geographica Sinica, 2022, 77(9): 2280−2291 [24] HARLE K J, HOWDEN S M, HUNT L P, et al. The potential impact of climate change on the Australian wool industry by 2030[J]. Agricultural Systems, 2007, 93(1/2/3): 61−89 [25] DOCKERTY T, LOVETT A, APPLETON K, et al. Developing scenarios and visualisations to illustrate potential policy and climatic influences on future agricultural landscapes[J]. Agriculture, Ecosystems & Environment, 2006, 114(1): 103−120 [26] HOBBS S E. Climate Change 1992: the supplementary report to the IPCC scientific assessment[J]. Journal of Atmospheric and Terrestrial Physics, 1996, 58(10): 1189 [27] HOUGHTON J, JENKINS G, EPHRAUMS J J. Climate change: the IPCC scientific assessment[M]. Cambridge: Cambridge University Press, 1990 [28] SYMONDS M. Faculty opinions recommendation of IPCC, 2021: summary for policymakers[M]//MASSON-DELMOTTE V, ZHAI P, PIRANI A, et al. Climate Change 2021: the Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge: Cambridge University Press, 2021 [29] VAN VUUREN D P, RIAHI K, MOSS R, et al. A proposal for a new scenario framework to support research and assessment in different climate research communities[J]. Global Environmental Change, 2012, 22(1): 21−35 doi: 10.1016/j.gloenvcha.2011.08.002 [30] SALINGER M J, STIGTER C J, DAS H P. Agrometeorological adaptation strategies to increasing climate variability and climate change[J]. Agricultural and Forest Meteorology, 2000, 103(1/2): 167−184 [31] VAN KLOMPENBURG T, KASSAHUN A, CATAL C. Crop yield prediction using machine learning: a systematic literature review[J]. Computers and Electronics in Agriculture, 2020, 177: 105709 doi: 10.1016/j.compag.2020.105709 [32] GONZALEZ-SANCHEZ A, FRAUSTO-SOLIS J, OJEDA-BUSTAMANTE W. Attribute selection impact on linear and nonlinear regression models for crop yield prediction[J]. The Scientific World Journal, 2014, 2014: 509429 [33] FILIPPI P, JONES E J, WIMALATHUNGE N S, et al. An approach to forecast grain crop yield using multi-layered, multi-farm data sets and machine learning[J]. Precision Agriculture, 2019, 20(5): 1015−1029 doi: 10.1007/s11119-018-09628-4 [34] KHANAL S, KUSHAL K C, FULTON J P, et al. Remote sensing in agriculture—accomplishments, limitations, and opportunities[J]. Remote Sensing, 2020, 12(22): 3783 doi: 10.3390/rs12223783 [35] DOROSH P A, RASHID S. Food and Agriculture in Ethiopia: Progress and Policy Challenges[M]. University of Pennsylvania Press, 2012 [36] HAILE S. Population, development, and environment in Ethiopia[J]. Environmental Change and Security Project Report, 2004, 10: 43−51 [37] EGZIABHER T B G. Vegetation and environment of the mountains of Ethiopia: implications for utilization and conservation[J]. Mountain Research and Development, 1988, 8(2/3): 211 doi: 10.2307/3673449 [38] ETEFA G, FRANKL A, LANCKRIET S, et al. Changes in land use/cover mapped over 80 years in the Highlands of Northern Ethiopia[J]. Journal of Geographical Sciences, 2018, 28(10): 1538−1559 doi: 10.1007/s11442-018-1560-3 [39] WILBY R L. Non-stationarity in daily precipitation series: implications for gcm down-scaling using atmospheric circulation indices[J]. International Journal of Climatology, 1997, 17(4): 439−454 doi: 10.1002/(SICI)1097-0088(19970330)17:4<439::AID-JOC145>3.0.CO;2-U [40] LEMBRECHTS J J, VAN-DEN-HOOGEN J, AALTO J, et al. Global maps of soil temperature[J]. Global Change Biology, 2022, 28(9): 3110−44 doi: 10.1111/gcb.16060 [41] LIU C L, CHEN N, LONG J C, et al. Review of the observed energy flow in the earth system[J]. Atmosphere, 2022, 13(10): 1738 doi: 10.3390/atmos13101738 [42] HENGL T, MENDES DE JESUS J, HEUVELINK G B M, et al. SoilGrids250m: global gridded soil information based on machine learning[J]. PLoS One, 2017, 12(2): e0169748 doi: 10.1371/journal.pone.0169748 [43] CISCAR J C, FISHER-VANDEN K, LOBELL D B. Synthesis and Review: an inter-method comparison of climate change impacts on agriculture[J]. Environmental Research Letters, 2018, 13(7): 070401 doi: 10.1088/1748-9326/aac7cb [44] KULKARNI S, DEO M C, GHOSH S. Evaluation of wind extremes and wind potential under changing climate for Indian offshore using ensemble of 10 GCMs[J]. Ocean & Coastal Management, 2016, 121: 141−152 [45] RÄISÄNEN J. How reliable are climate models?[J]. Tellus A:Dynamic Meteorology and Oceanography, 2007, 59(1): 2 doi: 10.1111/j.1600-0870.2006.00211.x [46] SCHMIDT G A. Enhancing the relevance of palaeoclimate model/data comparisons for assessments of future climate change[J]. Journal of Quaternary Science, 2010, 25(1): 79−87 doi: 10.1002/jqs.1314 [47] TEBALDI C, KNUTTI R. The use of the multi-model ensemble in probabilistic climate projections[J]. Philosophical Transactions Series A, Mathematical, Physical, and Engineering Sciences, 2007, 365(1857): 2053−2075 [48] ALFRED R, OBIT J H, CHIN C P Y, et al. Towards paddy rice smart farming: a review on big data, machine learning, and rice production tasks[J]. IEEE Access, 2021, 9: 50358−50380 doi: 10.1109/ACCESS.2021.3069449 [49] REHMAN T U, MAHMUD M S, CHANG Y K, et al. Current and future applications of statistical machine learning algorithms for agricultural machine vision systems[J]. Computers and Electronics in Agriculture, 2019, 156: 585−605 doi: 10.1016/j.compag.2018.12.006 [50] SHARMA A, JAIN A, GUPTA P, et al. Machine learning applications for precision agriculture: a comprehensive review[J]. IEEE Access, 2020, 9: 4843−4873 [51] GANAIE M A, HU M H, MALIK A K, et al. Ensemble deep learning: a review[J]. Engineering Applications of Artificial Intelligence, 2022, 115: 105151 doi: 10.1016/j.engappai.2022.105151 [52] JOUBERT A M, HEWITSON B C. Simulating present and future climates of southern Africa using general circulation models[J]. Progress in Physical Geography:Earth and Environment, 1997, 21(1): 51−78 doi: 10.1177/030913339702100104 [53] ZHOU T J, CHEN X L, WU B, et al. A robustness analysis of CMIP5 models over the east asia-western North Pacific domain[J]. Engineering, 2017, 3(5): 773−778 doi: 10.1016/J.ENG.2017.05.018 [54] VON SCHNEIDEMESSER E, DRISCOLL C, RIEDER H E, et al. How will air quality effects on human health, crops and ecosystems change in the future?[J]. Philosophical Transactions Series A, Mathematical, Physical, and Engineering Sciences, 2020, 378(2183): 20190330 [55] DI SALVO C. Improving results of existing groundwater numerical models using machine learning techniques: a review[J]. Water, 2022, 14(15): 2307 doi: 10.3390/w14152307 [56] HAILE G G, TANG Q, SUN S, et al. Droughts in East Africa: Causes, impacts and resilience[J]. Earth-science reviews. 2019, 193: 146-61. [57] FISCHER S, HILGER T, PIEPHO H P, et al. Do we need more drought for better nutrition? The effect of precipitation on nutrient concentration in East African food crops[J]. Science of the Total Environment. 2019, 658: 405-15. -