Hyperspectral remote sensing inversion of meadow aboveground biomass based on an XGBoost algorithm

doi:10.11686/cyxb2020190

Abstract

Abstract:

It is of great significance to agricultural production and animal husbandry， the management of grassland resources， and the sustainable utilization of forage to have efficient access to accurate information on aboveground biomass. This paper uses field data on the spectral reflectance of the grass canopy together with data on the aboveground biomass obtained in the same period， to analyze the correlation between spectral differences and meadow aboveground biomass， using a mutual information method. The correlation between the optimized vegetation index and the aboveground biomass of the meadow is then analyzed. In this paper， an extreme gradient boosting （XGBoost） algorithm and a simulation model estimating the aboveground biomass of different vegetation spectral index data sets are constructed. This model is compared with a model established by multiple linear regression （MLR） and random forest （RF） methodology. It was found that the first-order and second-order differentiation of spectral reflectance curves and the conjoint application of spectral vegetation index transformation from both spectral differentiation orders is helpful to improve the correlation between canopy spectra and aboveground biomass. Of the three approaches tested using the original spectral vegetation index to predict aboveground biomass， the XGBoost algorithm simulation model performed best， followed by the RF model. The MLR model displayed poor accuracy. The root mean square error of the XGBoost model was 140.26 g·m^-2； the mean absolute error was 97.20 g·m^-2； the Nash-Sutcliffe efficiency coefficient was 0.81， and the index of agreement was 0.94. It is concluded that the XGBoost algorithm can be used to create models for the estimation of steppe meadow aboveground biomass. The technology developed in this study provides a method for the rapid and accurate hyperspectral remote sensing monitoring of forage biomass， and lays a foundation for high-precision estimation of regional grassland herbage accumulation on a regional scale.

Key words: aboveground biomass, extreme gradient boosting algorithm, hyperspectral reflectance, vegetation index

Yi-ran ZHANG, Ting-xi LIU, Xin TONG, Li-min DUAN, Yu-chen WU. Hyperspectral remote sensing inversion of meadow aboveground biomass based on an XGBoost algorithm[J]. Acta Prataculturae Sinica, 2021, 30(4): 1-12.

Figures/Tables 9

Fig.1 Location of the study area and the test sites

Table 1 Vegetation indexes and related formula

植被指数Vegetation index	计算公式Expression	文献Reference
归一化植被指数 Normalized difference vegetation index (NDVI)	$N D V I = N I R - R N I R + R$	[19]
比值植被指数 Ratio vegetation index (RVI)	$R V I = N I R R$	[20]
土壤调节植被指数 Soil adjusted vegetation index (SAVI)	$S A V I = 1.5 N I R - R N I R + R + 0.5$	[21]
优化土壤调节植被指数 Optimization of soil-adjusted vegetation index (OSAVI)	$O S A V I = 1 + 0.16 N I R - R N I R + R + 0.16$	[22]
增强型植被指数2 Enhanced vegetation index 2 (EVI2)	$E V I 2 = 2.5 N I R - R N I R + 2.4 R + 1$	[23]
重归一化植被指数 Renormalized difference vegetation index (RDVI)	$R D V I = N I R - R N I R + R$	[24]
修改型土壤调节植被指数 Modified soil adjusted vegetation index (MSAVI)	$M S A V I = 12 × 2 N I R + 1 - 2 N I R + 1 2 - 8 N I R - R$	[25]

Table 1 Vegetation indexes and related formula

植被指数Vegetation index	计算公式Expression	文献Reference
归一化植被指数 Normalized difference vegetation index (NDVI)	$N D V I = N I R - R N I R + R$	[19]
比值植被指数 Ratio vegetation index (RVI)	$R V I = N I R R$	[20]
土壤调节植被指数 Soil adjusted vegetation index (SAVI)	$S A V I = 1.5 N I R - R N I R + R + 0.5$	[21]
优化土壤调节植被指数 Optimization of soil-adjusted vegetation index (OSAVI)	$O S A V I = 1 + 0.16 N I R - R N I R + R + 0.16$	[22]
增强型植被指数2 Enhanced vegetation index 2 (EVI2)	$E V I 2 = 2.5 N I R - R N I R + 2.4 R + 1$	[23]
重归一化植被指数 Renormalized difference vegetation index (RDVI)	$R D V I = N I R - R N I R + R$	[24]
修改型土壤调节植被指数 Modified soil adjusted vegetation index (MSAVI)	$M S A V I = 12 × 2 N I R + 1 - 2 N I R + 1 2 - 8 N I R - R$	[25]

Fig.2 Learning curves

Fig.3 Correlation analysis of the hyperspectral reflectance and AGB using the mutual information method

Table 2 Optimal vegetation indexes against AGB and the corresponding band combinations

光谱指数 Spectral index	基于原始光谱计算 Calculation based on raw spectrum			基于一阶微分光谱计算 Calculation based on first-order differential spectrum			基于二阶微分光谱计算 Calculation based on second-order differential spectrum
光谱指数 Spectral index	波段1 Band 1	波段2 Band 2	互信息度 MI	波段1 Band 1	波段2 Band 2	互信息度 MI	波段1 Band 1	波段2 Band 2	互信息度 MI
NDVI	748RE	752NIR	0.43	533G	751NIR	0.55	549G	747RE	0.39
RVI	748RE	752NIR	0.43	533G	751NIR	0.53	562G	744RE	0.43
SAVI	741RE	760NIR	0.42	527G	751NIR	0.56	530G	740RE	0.47
OSAVI	738RE	760NIR	0.43	527G	751NIR	0.56	530G	740RE	0.47
EVI2	769NIR	746RE	0.38	527G	751NIR	0.56	530G	740RE	0.47
RDVI	749RE	767NIR	0.40	521G	751NIR	0.58	447B	704RE	0.36
MSAVI	750RE	751NIR	0.39	527G	751NIR	0.56	530G	740RE	0.48

Fig.4 Correlation matrix plots of the original all available wavebands vegetation indexes and aboveground biomass (AGB) using the mutual information method

Fig.5 Correlation matrix plots of the first-order differential all available wavebands vegetation indexes and AGB using the mutual information method

Fig.6 Correlation matrix plots of the second-order differential all available wavebands vegetation indexes and AGB using the mutual information method

Fig.7 Results of each model with related accuracy parameters

References 37

1	Li W， Dou Z G， Wang Y， et al. Estimation of above-ground biomass of reed （Phragmites communis） based on in situ hyperspectral data in Beijing Hanshiqiao Wetland， China. Wetlands Ecology and Management， 2019， 27（1）： 87-102.
2	Anaya J A， Chuvieco E， Orueta A P. Aboveground biomass assessment in Colombia： A remote sensing approach. Forest Ecology and Management， 2008， 257（4）： 1237-1246.
3	Fava F， Colombo S， Bocchi S， et al. Identification of hyperspectral vegetation indices for Mediterranean pasture characterization. International Journal of Applied Earth Observations and Geoinformation， 2009， 11（4）： 233-243.
4	Wei X H， Jin G L， Fan Y M， et al. Estimation of biomass based on HJ-HIS data in Seriphidium transiliense desert grassland. Pratacultural Science， 2017， 34（12）： 2504-2511.
	魏秀红，靳瑰丽，范燕敏，等. 基于HJ-HSI数据的伊犁绢蒿荒漠草地生物量估测. 草业科学， 2017， 34（12）： 2504-2511.
5	Wang J， Brown D G， Bai Y. Investigating the spectral and ecological characteristics of grassland communities across an ecological gradient of the Inner Mongolian grasslands with in situ hyperspectral data. International Journal of Remote Sensing， 2014， 35（20）： 7179-7198.
6	Lv X D， Wang J G， Sun Q Z， et al. Research on the hyperspectral remote sensing estimation models for the fresh yield of alfalfa grassland. Acta Prataculturae Sinica， 2014， 23（1）： 84-91.
	吕小东，王建光，孙启忠，等. 苜蓿人工草地高光谱遥感估产模型的研究. 草业学报， 2014， 23（1）： 84-91.
7	Yang Y C， Zhao Y J， Qin K， et al. Prediction of black soil nutrient content based on airborne hyperspectral remote sensing. Transactions of the Chinese Society of Agricultural Engineering， 2019， 35（20）： 94-101.
	杨越超，赵英俊，秦凯，等. 黑土养分含量的航空高光谱遥感预测. 农业工程学报， 2019， 35（20）： 94-101.
8	Foster A J， Kakani V G， Mosali J. Estimation of bioenergy crop yield and N status by hyperspectral canopy reflectance and partial least square regression. Precision Agriculture， 2017， 18（2）： 192-209.
9	Wang C J， Jiang H H， Yin X J， et al. Evaluation for natural grassland utilization intensity based on GPS and UAV remote sensing for grassland biomass inversion. Transactions of the Chinese Society of Agricultural Engineering， 2018， 34（19）： 82-87.
	汪传建，江红红，尹小君，等. 基于GPS与无人机遥感反演草地生物量的放牧场利用强度评估. 农业工程学报， 2018， 34（19）： 82-87.
10	Ren H， Zhou G， Zhang X. Estimation of green aboveground biomass of desert steppe in Inner Mongolia based on red-edge reflectance curve area method. Biosystems Engineering， 2011， 109（4）： 385-395.
11	An H B， Li F， Zhao M L， et al. Optimized spectral indices based estimation of forage grass biomass. Spectroscopy and Spectral Analysis， 2015， 35（11）： 3155-3160.
	安海波，李斐，赵萌莉，等. 基于优化光谱指数的牧草生物量估算. 光谱学与光谱分析， 2015， 35（11）： 3155-3160.
12	Tong X， Duan L M， Liu T X， et al. Combined use of in situ hyperspectral vegetation indices for estimating pasture biomass at peak productive period for harvest decision. Precision Agriculture， 2019， 20（3）： 477-495.
13	Fava F， Parolo G， Colombo R， et al. Fine-scale assessment of hay meadow productivity and plant diversity in the European Alps using field spectrometric data. Agriculture Ecosystems and Environment， 2010， 137（1）： 151-157.
14	Chen T Q， Guestrin C. Xgboost： A scalable tree boosting system//Proceedings of the 22nd acm sigkdd internation conference on knowledge discovery and data mining-KDD’16. San Francisco： ACM Press the 22nd ACM SIGKDD International Conference， 2016： 785-794.
15	Mutanga O， Skidmore A K. Narrow band vegetation indices overcome the saturation problem in biomass estimation. International Journal of Remote Sensing， 2004， 25（19）： 3999-4014.
16	Behmann J， Steinrucken J， Plumer L. Detection of early plant stress responses in hyperspectral images. ISPRS Journal of Photogrammetry and Remote Sensing， 2014， 93： 98-111.
17	Cao W M， Liu X Y， Wang G L， et al. Combined analyses of MRT and DCCA on relationships between plant community distri-bution and ecological factors of Horqin Sandy Land. Chinese Journal of Ecology， 2017， 36（2）： 318-327.
	曹文梅，刘小燕，王冠丽，等. 科尔沁沙地自然植被与生境因子的MRT分类及DCCA分析. 生态学杂志， 2017， 36（2）： 318-327.
18	Fraser A M， Swinney H L. Independent coordinates for strange attractors from mutual information. Physical Review A， 1986， 33（2）： 1134-1140.
19	Tucker C J. Red and photographic infrared linear combinations for monitoring vegetation. Remote Sensing and Environment， 1979， 8（2）： 127-150.
20	Jordan C F. Derivation of leaf-area index from quality of light on the forest floor. Ecology， 1969， 50（4）： 663-666.
21	Huete A R. A soil-adjusted vegetation index （SAVI）. Remote Sensing of Environment， 1988， 25（3）： 295-309.
22	Rondeaux G， Steven M， Baret F. Optimization of soil-adjusted vegetation indices. Remote Sensing of Environment， 1996， 55（2）： 95-107.
23	Jiang Z， Huete A R， Didan K， et al. Development of a two-band enhanced vegetation index without a blue band. Remote Sensing of Environment， 2008， 112（10）： 3833-3845.
24	Roujean J L， Breon F M. Estimating PAR absorbed by vegetation from bidirectional reflectance measurements. Remote Sensing of Environment， 1995， 51（3）： 375-384.
25	Qi J G， Chehbouni A R， Huete A R， et al. A modified soil adjusted vegetation index. Remote Sensing of Environment， 1994， 48（2）： 119-126.
26	Breiman L. Random forest. Machine Learning， 2001， 45（1）： 5-32.
27	Nash J E， Sutcliffe J V. River flow forecasting through conceptual models part I — A discussion of principles. Journal of Hydrology， 1970， 10（3）： 282-290.
28	Willmott C J， Ackleson S G， Davis R E， et al. Statistics for the evaluation and comparison of models. Journal of Geophysical Research： Earth Surface， 1985， 90（5）： 8995-9005.
29	Wu H Q， Fan Y M， Jin G L， et al. Spectral characteristics of main plant in Seriphidium transiliense desert grassland. Pratacultural Science， 2019， 36（7）： 1765-1773.
	武红旗，范燕敏，靳瑰丽，等. 伊犁绢蒿荒漠草地植物光谱特征. 草业科学， 2019， 36（7）： 1765-1773.
30	Wang X P， Zhang F， Ding J L. Estimation of soil salt content （SSC） in the Ebinur Lake Wetland National Nature Reserve （ELWNNR）， Northwest China， based on a Bootstrap-BP neural network model and optimal spectral indices. Science of the Total Environment， 2018， 615： 918-930.
31	Fatehi P， Damm A， Schweiger A K， et al. Mapping alpine aboveground biomass from imaging spectrometer data： A comparison of two approaches. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing， 2015， 8（6）： 3123-3139.
32	Marshall M， Thenkabail P. Advantage of hyperspectral EO-1 Hyperion over multispectral IKONOS， GeoEye-1， WorldView-2， Landsat ETM+， and MODIS vegetation indices in crop biomass estimation. ISPRS Journal of Photogrammetry and Remote Sensing， 2015， 108： 205-218.
33	Foster A J， Kakani V G， Mosali J. Estimation of bioenergy crop yield and N status by hyperspectral canopy reflectance and partial least square regression. Precision Agriculture， 2017， 18（2）： 192-209.
34	Delegido J， Fernandez G， Gandia S， et al. Retrieval of chlorophyll content and LAI of crops using hyperspectral techniques： Application to PROBA/CHRIS data. International Journal of Remote Sensing， 2008， 29（24）： 7107-7127.
35	Hansen P M， Schjoerring J K. Reflectance measurement of canopy biomass and nitrogen status in wheat crops using normalized difference vegetation indices and partial least squares regression. Remote Sensing of Environment， 2003， 86： 542-553.
36	Jin X L， Li Z H， Feng H K. Deep neural network algorithm for estimating maize biomass based on simulated Sentinel 2A vegetation indices and leaf area index. Journal of Citation Reports， 2020， 8（1）： 87-97.
37	Wang L， Chang Q R， Yang J， et al. Estimation of paddy rice leaf area index using machine learning methods based on hyperspectral data from multi-year experiments. PLoS One， 2018， 13（12）： e0207624.

[1]	Jia-meng DU, Gang BAO, Si-qin TONG, Xiao-jun HUANG, Wendurina, Meili, Yu-hai BAO. Variations in vegetation cover and its relationship with climate change and human activities in Mongolia during the period 1982-2015 [J]. Acta Prataculturae Sinica, 2021, 30(2): 1-13.
[2]	Hui-fang ZHAO, Xiao-dong LI, Dong ZHANG, Rui-xiang XIAO. Aboveground biomass in grasslands in Qinghai Province estimated from MODIS data and its influencing factors [J]. Acta Prataculturae Sinica, 2020, 29(12): 5-16.
[3]	WANG Chun-yan, YAN Xia, GU Meng-he. The interaction effects of vegetation and soil nutrients on vegetation succession in abandoned farmland on the Loess Plateau [J]. Acta Prataculturae Sinica, 2018, 27(11): 26-35.
[4]	ZHAO Jie, LI Wei, JING Guang-Hua, WEI Lin, CHENG Ji-Min. Responses of species diversity and aboveground biomass to nitrogen addition in fenced and grazed grassland on the Loess Plateau [J]. Acta Prataculturae Sinica, 2017, 26(8): 54-64.
[5]	GE Jing, MENG Bao-Ping, YANG Shu-Xia, GAO Jin-Long, YIN Jian-Peng, ZHANG Ren-Ping, FENG Qi-Sheng, LIANG Tian-Gang. Monitoring of above-ground biomass in alpine grassland based on agricultural digital camera and MODIS remote sensing data: A case study in the Yellow River Headwater Region [J]. Acta Prataculturae Sinica, 2017, 26(7): 23-34.
[6]	JING Mei-Ling, MA Yu-Shou, LI Shi-Xiong, WANG Yan-Long. Initial assessment of sixteen perennial grass species in the upper Datong River [J]. Acta Prataculturae Sinica, 2017, 26(6): 76-88.
[7]	SUN Yong-Xiu, YAN Cheng, XU Hai-Liang, YAO Yan-Li. Grassland community species diversity and aboveground biomass responses to difference in cover soil thickness in restoration after mining damage [J]. Acta Prataculturae Sinica, 2017, 26(1): 54-62.
[8]	YANG Shu-Xia, ZHANG Wen-Juan, FENG Qi-Sheng, MENG Bao-Ping, GAO Jin-Long, LIANG Tian-Gang. Monitoring of grassland herbage accumulation by remote sensing using MODIS daily surface reflectance data in the Qingnan Region [J]. Acta Prataculturae Sinica, 2016, 25(8): 14-26.
[9]	WANG Lu, DING Jian-Li. Vegetation index feature change and its influencing factors and spatial-temporal process analysis of desert grassland in the Ebinur Lake Nature Reserve, Xinjiang [J]. Acta Prataculturae Sinica, 2015, 24(5): 4-11.
[10]	ZHOU Xin,ZUO Xiao-an,ZHAO Xue-yong,WANG Shao-kun,LUO Yong-qing,YUE Xiang-fei,ZHANG La-mei. Effect of change in semiarid sand dune habitat on aboveground plant biomass, carbon and nitrogen [J]. Acta Prataculturae Sinica, 2014, 23(6): 36-44.
[11]	LI Qiang,ZHOU Dao-wei,SONG Yan-tao. The distribution features of two widespread legumes and their relationships with soil factors in Songnen grassland [J]. Acta Prataculturae Sinica, 2014, 23(1): 31-40.
[12]	CHEN Meng-die, HUANG Xiao-dong, HOU Xiu-min, FENG Qi-sheng, YU Hui, GUO Zheng-gang, LIANG Tian-gang. Dynamic monitoring of biomass and vegetation coverage in rodent damaged grassland regions of Qinghai Province, China [J]. Acta Prataculturae Sinica, 2013, 22(4): 247-256.
[13]	QIAN Yu-rong, YANG Feng, YU Jiong, JIA Zhen-hong, LI Jian-long, Palidan Tuerxun. Vegetation index feature and spatial-temporal process analysis of desert grassland in the Fukang area of Xinjiang [J]. Acta Prataculturae Sinica, 2013, 22(3): 25-.
[14]	ZHOU Yu-ting, FU Gang, SHEN Zhen-xi, ZHANG Xian-zhou, WU Jian-shuang, LI Yun-long, YANG Peng-wan. Estimation model of aboveground biomass in the Northern Tibet Plateau based on remote sensing date [J]. Acta Prataculturae Sinica, 2013, 22(1): 120-129.
[15]	QIAN Yu-rong, YU Jiong, JIA Zhen-hong, YANG Feng, Palidan Tuerxun. Extraction and analysis of hyper-spectral data from typical desert grassland in Xinjiang [J]. Acta Prataculturae Sinica, 2013, 22(1): 157-166.