哈萨克斯坦北部小麦遥感估产方法研究

doi:10.12118/j.issn.1000–6060.2021.236

摘要/Abstract

摘要：

以哈萨克斯坦北部雨养耕作区为研究靶区,基于春小麦产量统计数据和遥感光谱指数,开展了春小麦估产最优预测时期及植被指数分析,采用回归分析、随机森林、支持向量机及双向循环神经网络模型估算春小麦产量,并对比分析了不同模型的模拟精度。结果表明：北哈萨克斯坦州、阿克莫拉州和库斯塔纳州2007—2016年春小麦估产的最佳预测时期为6月26日—8月5日,该时期是春小麦产量形成的关键时期。北哈萨克斯坦州春小麦估产最优植被指数为7月12日的绿度叶绿素指数（Green chlorophyll index, CI_green）,阿克莫拉州春小麦估产最优植被指数为8月5日的绿度动态宽波段指数（Green wide dynamic range vegetation index, WDRVI_green）,库斯塔纳州春小麦最优估产植被指数为7月12日的WDRVI_green。对比分析4种模型模拟春小麦产量的精度,在样本点较少的情况下,双向循环神经网络模型相比其他模型在估算哈萨克斯坦北部三州春小麦产量上精度较高;春小麦产量与植被净初级生产力NPP相关性分析结果显示,北哈萨克斯坦州、阿克莫拉州和库斯塔纳州决定系数R²在0.50以上面积占比分别为44%、94%和77%,表明上述估产模型可应用于哈萨克斯坦北部三州春小麦估产,尤其是阿克莫拉州和库斯塔纳州。

关键词: 雨养小麦耕作区, 遥感估产, 植被指数, 回归模型, 机器学习, 哈萨克斯坦北部

Abstract:

Kazakhstan’s flour export volume ranks first in the world and is called the Central Asian granary. Its northern regions, including North Kazakhstan, Aqmola, and Qostanay, are the world’s important wheat and flour exporters. The proportion of wheat planting structures has reached 86%. Since 2010, its wheat and barley output has ranked 12^th in the world, and its export volume has ranked 5^th in the world. However, the region is rain-fed and lacks effective irrigation measures. It often suffers from drought stress due to its location in the monsoon climatic zone, resulting in a large-scale reduction in spring wheat production, which severely restricts the economic development of countries that rely on wheat imports. The estimation of wheat production plays a vital crucial role in promoting regional food security and social stability, especially in responding to the food crisis in the post-epidemic era and achieving zero hunger advocated by the United Nations. In this paper, the rain-fed farming area in northern Kazakhstan is used as the research target area. An analysis of the optimal vegetation index for spring wheat yield estimation was conducted based on the statistical spring wheat yield and vegetation index. Various methods, such as regression, random forest, support vector machine, and neural network, are evaluated for the accuracy of wheat yield estimation. In North Kazakhstan, the best vegetation index for estimating spring wheat yield is the greenness chlorophyll index from 2007 to 2016. The forecast time can be advanced to July 12^th from 2007 to 2016. In Aqmola, the best vegetation index for estimating spring wheat yield is the greenness dynamic wide-band vegetation index (WDRVI_green). The forecast time can be advanced to August 5^th. In Qostanay, the best vegetation index for estimating spring wheat yield is the WDRVI_green. The forecast time can be advanced to July 12^th. In this paper, MODIS NPP products are selected to verify the estimation results. Through correlation analysis with NPP shows that the neural network has higher accuracy in estimating spring wheat yield in the three northern states of Kazakhstan than other models. Neural networks accounted for 44%, 94%, and 77% of correlations in North Kazakhstan, Aqmola, and Qostanay states, respectively.

Key words: rain-fed wheat farming area, remote sensing yield estimation, vegetation index, regression model, machine learning, northern Kazakhstan

尹瀚民,古丽·加帕尔,于涛,Jeanine UMUHOZA,李旭. 哈萨克斯坦北部小麦遥感估产方法研究[J]. 干旱区地理, 2022, 45(2): 488-498.

YIN Hanmin,Guli JIAPAER,YU Tao,Jeanine UMUHOZA,LI Xu. Wheat yield estimation with remote sensing in northern Kazakhstan[J]. Arid Land Geography, 2022, 45(2): 488-498.

图/表 7

图1

表1

植被指数公式"

植被指数	公式	参考文献
NDVI	$NDVI = (ρ 2 - ρ 1) ∕ (ρ 2 + ρ 1)$	[28-30]
EVI2	$EVI 2 = [2.5 × ρ 2 - ρ 1] ∕ (ρ 2 + 2.4 × ρ 1 + 1)$	[31]
WDRVI	$WDRVI = (α × ρ 2 - ρ 1) (α × ρ 2 + ρ 1), α$ 取值为0.1	[32]
SANDVI	$SANDVI = NDVI, NDVI ≤ 0.78 0.03 × RVI + 0.5363, NDVI > 0.78$ $RVI = ρ 2 / ρ 1$	[33]
WDRVI_green	$WDRV I green = [(α × ρ 2 - ρ 4) / (α × ρ 2 + ρ 4)] + [(1 - α) / (1 + α)], α 取值为 0.3$	[34]
CI_green	$C I green = (ρ 2 / ρ 4) - 1$	[35]
DVI	$DVI = ρ 2 - ρ 1$	[36]
OSAVI	$OSAVI = 1 + 1.16 × [(ρ 2 - ρ 1) ∕ (ρ 2 + ρ 1 + 0.16)$ ]	[37]
NMDI	$NMDI = [ρ 2 - (ρ 6 - ρ 7)] ∕ [ρ 2 + (ρ 6 - ρ 7)]$	[8]

表1

图2

表2

图3

图4

图5

参考文献 41

[1]	李宁. 哈萨克斯坦小麦产业的发展与问题[J]. 粮食流通技术, 2013(2):37-42.
	[ Li Ning. Study on Kazakhstan’s wheat industry and its problems[J]. Grain Distribution Technology, 2013(2):37-42. ]
[2]	马骏, 龚新蜀. 中亚国家粮食安全问题研究[J]. 世界农业, 2014(8):22-26.
	[ Ma Jun, Gong Xinshu. Research on food security in Central Asian countries[J]. World Agriculture, 2014(8):22-26. ]
[3]	周清波. 国内外农情遥感现状与发展趋势[J]. 中国农业资源与区划, 2004, 25(5):12-17.
	[ Zhou Qingbo. Status and tendency for development in remote sensing of agriculture situation[J]. Journal of China Agricultural Resources and Regional Planning, 2004, 25(5):12-17. ]
[4]	Anup K P, Lim C, Ramesh P S, et al. Crop yield estimation model for Iowa using remote sensing and surface parameters[J]. International Journal of Applied Earth Observation and Geoinformation, 2006, 8(1):26-33. doi: 10.1016/j.jag.2005.06.002
[5]	Michael S R. Assessment of millet yields and production in northern Burkina Faso using integrated NDVI from the AVHRR[J]. International Journal of Remote Sensing, 1992, 13(18):3871-3879.
[6]	Michael S R. Operational yield forecast using AVHRR NDVI data: Reduction of environmental and inter-annual variability[J]. International Journal of Remote Sensing, 1997, 18(5):1059-1077. doi: 10.1080/014311697218575
[7]	焦险峰, 杨邦杰, 裴志远, 等. 基于植被指数的作物产量监测方法研究[J]. 农业工程学报, 2005, 21(4):104-108.
	[ Jiao Xianfeng, Yang Bangjie, Pei Zhiyuan, et al. Monitoring crop yield using NOAA/AVHRR based vegetation indices[J]. Transactions of the Chinese Society of Agricultural Engineering, 2005, 21(4):104-108. ]
[8]	Tuvdendorj B. Determination of appropriate remote sensing indices for spring wheat yield estimation in Mongolia[J]. Remote Sensing, 2019, 11(21):2568-2570. doi: 10.3390/rs11212568
[9]	李军玲, 郭其乐, 彭记永. 基于MODIS数据的河南省冬小麦产量遥感估算模型[J]. 生态环境学报, 2012, 21(10):1665-1669.
	[ Li Junling, Guo Qile, Peng Jiyong. Remote sensing estimation model of Henan Province winter wheat yield based on MODIS data[J]. Ecology and Environmental Sciences, 2012, 21(10):1665-1669. ]
[10]	谭昌伟. 基于开花期卫星遥感数据的大田小麦估产方法比较[J]. 中国农业科学, 2017, 50(16):3101-3109.
	[ Tan Changwei. Comparison of the methods for predicting wheat yield based on satellite remote sensing data at anjournal[J]. Scientia Agricultura Sinica, 2017, 50(16):3101-3109. ]
[11]	Uno Y. Artificial neural networks to predict corn yield from compact airborne spectrographic imager data[J]. Computers and Electronics in Agriculture, 2005, 47(2):149-161.
[12]	王恺宁, 王修信. 多植被指数组合的冬小麦遥感估产方法研究[J]. 干旱区资源与环境, 2017, 31(7):44-49.
	[ Wang Kaining, Wang Xiuxin. Research on winter wheat yield estimtion with the multipy remote sensing vegetation index combination[J]. Journal of Arid Land Resources and Environment, 2017, 31(7):44-49. ]
[13]	徐春萌. 基于DSSAT作物模型的中美大豆主产区单产模拟与验证[J]. 农业工程学报, 2021, 37(3):132-139.
	[ Xu Chunmeng. Simulations and validations of the soybean yields per unit area using DSSAT crop model in the major soybean producing areas of China and America[J]. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(3):132-139. ]
[14]	赵军. 基于净初级生产力的春小麦生产潜力及估产研究——以甘肃省白银区为例[J]. 干旱地区农业研究, 2015, 33(1):199-204.
	[ Zhao Jun. Potential productivity and yield estimation of spring wheat based on a net primary production model: Taking Baiyin District in Gansu Province as an example[J]. Agricultural Research in the Arid Areas, 2015, 33(1):199-204. ]
[15]	Bolton D K, Friedl M A. Forecasting crop yield using remotely sensed vegetation indices and crop phenology metrics[J]. Agricultural and Forest Meteorology, 2013, 173:74-84. doi: 10.1016/j.agrformet.2013.01.007
[16]	Leroux L. Crop monitoring using vegetation and thermal indices for yield estimates: Case study of a rainfed cereal in semi-arid west Africa[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016, 9(1):347-362. doi: 10.1109/JSTARS.4609443
[17]	郭锐. 山东省冬小麦单产监测与预报方法研究[J]. 农业机械学报, 2020, 51(7):156-163.
	[ Guo Rui. Monitoring and forecasting method of winter wheat yield in Shandong Province[J]. Transactions of the Chinese Society for Agricultural Machinery, 2020, 51(7):156-163. ]
[18]	安秦. 长春地区玉米产量遥感估算方法研究[D]. 长春: 吉林大学, 2018.
	[ An Qin. Research on methods of maize yield estimation by remote sensing in Changchun region[D]. Changchun: Jilin University, 2018. ]
[19]	曾妍, 王迪, 赵小娟, 等. 基于支持向量回归的关中平原冬小麦估产研究[J]. 中国农业信息, 2019, 31(6):10-20.
	[ Zeng Yan, Wang Di, Zhao Xiaojuan, et al. Study on yield prediction of winter wheat in Guanzhong Plain based on SVR[J]. China Agricultural Informatics, 2019, 31(6):10-20. ]
[20]	黄健熙, 马鸿元, 田丽燕, 等. 基于时间序列LAI和ET同化的冬小麦遥感估产方法比较[J]. 农业工程学报, 2015, 31(4):197-203.
	[ Huang Jianxi, Ma Hongyuan, Tian Liyan, et al. Comparison of remote sensing yield estimation methods for winter wheat based on assimilating time-sequence LAI and ET[J]. Transactions of the Chinese Society of Agricultural Engineering, 2015, 31(4):197-203. ]
[21]	马鸿元, 黄健熙, 黄海, 等. 基于历史气象资料和WOFOST模型的区域产量集合预报[J]. 农业机械学报, 2018, 49(9):257-266.
	[ Ma Hongyuan, Huang Jianxi, Huang Hai, et al. Ensemble forecasting of regional yield of winter wheat based on WOFOST model using historical metrological dataset[J]. Transactions of the Chinese Society for Agricultural Machinery, 2018, 49(9):257-266. ]
[22]	杨练兵, 郑宏伟, 罗格平, 等. 基于遗传算法优化BP神经网络的土壤盐渍化反演[J]. 地理与地理信息科学, 2021, 37(2):21-37.
	[ Yang Lianbing, Zheng Hongwei, Luo Geping, et al. Retrieval of soil salinity content based on BP neural network optimized by genetic algorithm[J]. Geography and Geo-Information Science, 2021, 37(2):21-37. ]
[23]	Dara A. Mapping the timing of cropland abandonment and recultivation in northern Kazakhstan using annual Landsat time series[J]. Remote Sensing of Environment, 2018, 213:49-60. doi: 10.1016/j.rse.2018.05.005
[24]	Guo H. Determining variable weights for an optimal scaled drought condition index (OSDCI): Evaluation in Central Asia[J]. Remote Sensing of Environment, 2019, 231:1-17.
[25]	Yin H. Monitoring fire regimes and assessing their driving factors in Central Asia[J]. Journal of Arid Land, 2021, 13(5):500-515. doi: 10.1007/s40333-021-0008-2
[26]	韩其飞, 罗格平, 白洁, 等. 基于多期数据集的中亚五国土地利用/覆盖变化分析[J]. 干旱区地理, 2012, 35(6):909-918.
	[ Han Qifei, Luo Geping, Bai Jie, et al. Characteristics of land use and cover change in Central Asia in recent 30 years[J]. Arid Land Geography, 2012, 35(6):909-918. ]
[27]	Yao F. Estimation of maize yield by using a process-based model and remote sensing data in the Northeast China Plain[J]. Physics and Chemistry of the Earth, 2015, 87:142-152.
[28]	Mkhabela M S, Mashinini N N. Early maize yield forecasting in the four agro-ecological regions of Swaziland using NDVI data derived from NOAA’s-AVHRR[J]. Agricultural and Forest Meteorology, 2005, 129(1):1-9. doi: 10.1016/j.agrformet.2004.12.006
[29]	田艳君, 石莹, 帅艳民, 等. 基于遥感时序特征的地表覆被信息提取[J]. 干旱区地理, 2021, 44(2):450-459.
	[ Tian Yanjun, Shi Ying, Shuai Yanmin, et al. Land cover information retrieval from temporal features based remote sensing images[J]. Arid Land Geography, 2021, 44(2):450-459. ]
[30]	王华, 杨乾鹏, 田云杰, 等. 基于多时相Landsat影像的中亚地区植被覆盖遥感监测[J]. 干旱区地理, 2020, 43(4):1023-1032.
	[ Wang Hua, Yang Qianpeng, Tian Yunjie, et al. Vegetation coverage monitoring in the Central Asian countries using multi-temporal Landsat images[J]. Arid Land Geography, 2020, 43(4):1023-1032. ]
[31]	Jiang Z. Development of a two-band enhanced vegetation index without a blue band[J]. Remote Sensing of Environment, 2008, 112(10):3833-3845. doi: 10.1016/j.rse.2008.06.006
[32]	Gitelson A A. Wide dynamic range vegetation index for remote quantification of biophysical characteristics of vegetation[J]. Journal of Plant Physiology, 2004, 161(2):165-173. pmid: 15022830
[33]	Gu Y. NDVI saturation adjustment: A new approach for improving cropland performance estimates in the Greater Platte River Basin, USA[J]. Ecological Indicators, 2013, 30:1-6. doi: 10.1016/j.ecolind.2013.01.041
[34]	Gitelson A A. Remote estimation of crop gross primary production with Landsat data[J]. Remote Sensing of Environment, 2012, 121:404-414. doi: 10.1016/j.rse.2012.02.017
[35]	Gitelson A A. Remote estimation of canopy chlorophyll content in crops[J]. Geophysical Research Letters, 2005, 32(8):1-4.
[36]	Mirchooli F. Spatial distribution dependency of soil organic carbon content to important environmental variables[J]. Ecological Indicators, 2020, 116:1-5.
[37]	Jin Y. Remote sensing-based biomass estimation and its spatio-temporal variations in temperate grassland, northern China[J]. Remote Sensing, 2014, 6(2):1496-1513. doi: 10.3390/rs6021496
[38]	高煜堃. 基于机器学习及多源数据的亚热带典型区域森林地上生物量估测研究[D]. 杭州: 浙江农林大学, 2018.
	[ Gao Yukun. Aboveground forest biomass estimation based on machine learning algorithms and multi-source data in a typical subtropical region[D]. Hangzhou: Zhejiang Agriculture and Forestry University, 2018. ]
[39]	郜燕芳, 李俊明, 刘东伟, 等. 基于随机森林模型的城市不透水面提取研究——以呼和浩特市为例[J]. 冰川冻土, 2018, 40(4):828-836.
	[ Gao Yanfang, Li Junming, Liu Dongwei, et al. Research on extraction of urban impervious surface based on random forest model: A case study in Hohhot[J]. Journal of Glaciology and Geocryology, 2018, 40(4):828-836. ]
[40]	徐新刚, 吴炳方, 蒙继华, 等. 农作物单产遥感估算模型研究进展[J]. 农业工程学报, 2008(2):290-298.
	[ Xu Xingang, Wu Bingfang, Meng Jihua, et al. Research advances in crop yield estimation models based on remote sensing[J]. Transactions of the Chinese Society of Agricultural Engineering, 2018(2):290-298. ]
[41]	徐逸, 董轩妍, 王俊杰. 4种机器学习模型反演太湖叶绿素a浓度的比较[J]. 水生态学杂志, 2019, 40(4):48-57.
	[ Xu Yi, Dong Xuanyan, Wang Junjie. Use of remote multispectral imaging to monitor chlorophyll-a in Taihu Lake: A comparison of machine learning models[J]. Journal of Hydroecology, 2019, 40(4):48-57. ]

地名	最优植被指数（排名前5位）	最佳观测日期（月-日）	R²	RMSE/kg
北哈萨克斯坦州	CI_green	07-12	0.83	131.8
	LAI	07-12	0.82	135.5
	LAI	06-26	0.80	143.0
	DVI	07-12	0.77	153.6
	OSAVI	07-12	0.76	155.3
阿克莫拉州	WDRVI_green	08-05	0.80	137.5
	EVI2	07-12	0.78	143.9
	LAI	07-12	0.77	145.0
	WDRVI	07-12	0.76	147.6
	EVI2	07-20	0.76	148.2
库斯塔纳州	WDRVI_green	07-12	0.88	123.6
	OSAVI	07-12	0.83	150.7
	CI_green	07-12	0.81	157.5
	DVI	07-12	0.81	158.2
	LAI	06-26	0.80	163.7