Retrieval of grassland aboveground biomass based on airborne LiDAR and SuperView-1 data

doi:10.11686/cyxb2022199

Abstract

Abstract:

Grassland aboveground biomass （AGB） is an important indicator in grassland monitoring. It is an important index when designing strategies for the ecological protection and rational utilization of grassland resources. In addition， it is of great significance for the sustainable development and scientific management of grassland. In this study， shrub grassland in Xing’an County， Guangxi was the subject of the research， and data were obtained from airborne LiDAR data and high-resolution multispectral satellite images. The retrieval of grassland AGB was investigated using data collected from 89 field quadrats in 2021 and five basic regression models. The accuracy of different indicators and models was evaluated by root mean square error （RMSE）， mean absolute error （MAE）， and R-square values. It was found that grass height metrics were very important information for grassland AGB retrieval. We calculated correlation coefficients between pairs of indexes. In terms of vegetation indexes， the highest correlation coefficient was between the enhanced vegetation index （EVI） and AGB （0.666）. In terms of vegetation height indexes， the highest correlation coefficient was between average grass height （CHM_mean） and AGB （0.686）. In terms of combined indexes， the highest correlation coefficient was between the ratio vegetation index （RVI）×CHM_mean and AGB （0.735）. The accuracy and verification results showed that the minimum RMSE of the EVI models was 292.047 g·m^-2， the minimum RMSE of CHM_mean models was 245.084 g·m^-2， and the minimum RMSE of RVI×CHM_mean models was 225.872 g·m^-2. Our results show that grass height information can be effectively extracted from airborne LiDAR data， and although there is an obvious underestimation， it still has great application potential in research on grassland AGB.

Key words: grassland aboveground biomass, airborne LiDAR, grass height, high-resolution satellite image, remotely sensed retrieval model

Kai-hong XU, Zhao SHI, Lei-chao MA, Ping WANG, Ang CHEN, Xing WANG, Ming CHENG, Yue-xin XIAO, Rong-tan WANG. Retrieval of grassland aboveground biomass based on airborne LiDAR and SuperView-1 data[J]. Acta Prataculturae Sinica, 2023, 32(5): 40-49.

Figures/Tables 9

Fig. 1 Location of the study area

Fig.2 Field quadrat

Table 1 Specifications of the RIEGL VUX-1UAV flight parameters

参数Parameter	值Value
测距精度Ranging accuracy	5 mm
最大脉冲频率Maximum pulse frequency	50 kHz
最大扫描频率Maximum line speed	200 lines·s^-1
视场角Field angle	330°
重量Weight	3.6 kg
体积Volume	227 mm×180 mm×125 mm
最大测距Maximum ranging	920 m

Fig.3 Altitude of point cloud

Fig.4 Scatter plots of LiDAR-derived and field measured grass height

Fig.5 Correlation analysis results between indexes and AGB

Table 2 Validation statistics of regression models for AGB estimation

自变量 Independent variables	模型 Models	公式 Formula	决定系数 R²	均方根误差 RMSE （g·m^-2）	平均绝对误差 MAE （g·m^-2）
增强型植被指数EVI	线性Linear	$y = 3341.6 x - 565.7$	0.417	292.556	219.446
	乘幂Power	$y = 3631.8272 x^{1.7156}$	0.416	295.184	214.888
	多项式Polynomial	$y = 2143.78 x - 81.64 x^{2} - 71.10 x^{3} + 809.24$	0.403	299.678	226.222
	对数Logarithmic	$y = 1320.4 l n ? x + 2009.0$	0.405	292.047	226.628
	指数Exponential	$y = 145.4094 e x p (4.0498 x)$	0.398	306.904	211.195
平均草层高度CHM_mean	线性Linear	$y = 3629.0 x + 384.4$	0.498	245.084	190.409
	乘幂Power	$y = 2548.3822 x^{0.5154}$	0.495	248.620	199.027
	多项式Polynomial	$y = 1798.86 x - 1011.58 x^{2} + 286.69 x^{3} + 809.24$	0.459	260.272	217.594
	对数Logarithmic	$y = 424.03 l n ? x + 1784.99$	0.473	254.714	208.202
	指数Exponential	$y = 542.0025 e x p (3.2795 x)$	0.486	261.219	202.727
比值植被指数×平均草层高度RVI×CHM_mean	线性Linear	$y = 412.29 x + 474.52$	0.587	225.872	147.937
	乘幂Power	$y = 949.7594 x^{0.41549}$	0.547	236.702	186.503
	多项式Polynomial	$y = 1948.08 x - 1366.34 x^{2} - 3.427 x^{3} + 809.243$	0.497	274.047	212.017
	对数Logarithmic	$y = 340.07 l n ? x + 981.64$	0.499	250.078	205.620
	指数Exponential	$y = 596.89456 e x p (0.35384 x)$	0.602	242.616	195.134

Table 2 Validation statistics of regression models for AGB estimation

自变量 Independent variables	模型 Models	公式 Formula	决定系数 R²	均方根误差 RMSE （g·m^-2）	平均绝对误差 MAE （g·m^-2）
增强型植被指数EVI	线性Linear	$y = 3341.6 x - 565.7$	0.417	292.556	219.446
	乘幂Power	$y = 3631.8272 x^{1.7156}$	0.416	295.184	214.888
	多项式Polynomial	$y = 2143.78 x - 81.64 x^{2} - 71.10 x^{3} + 809.24$	0.403	299.678	226.222
	对数Logarithmic	$y = 1320.4 l n ? x + 2009.0$	0.405	292.047	226.628
	指数Exponential	$y = 145.4094 e x p (4.0498 x)$	0.398	306.904	211.195
平均草层高度CHM_mean	线性Linear	$y = 3629.0 x + 384.4$	0.498	245.084	190.409
	乘幂Power	$y = 2548.3822 x^{0.5154}$	0.495	248.620	199.027
	多项式Polynomial	$y = 1798.86 x - 1011.58 x^{2} + 286.69 x^{3} + 809.24$	0.459	260.272	217.594
	对数Logarithmic	$y = 424.03 l n ? x + 1784.99$	0.473	254.714	208.202
	指数Exponential	$y = 542.0025 e x p (3.2795 x)$	0.486	261.219	202.727
比值植被指数×平均草层高度RVI×CHM_mean	线性Linear	$y = 412.29 x + 474.52$	0.587	225.872	147.937
	乘幂Power	$y = 949.7594 x^{0.41549}$	0.547	236.702	186.503
	多项式Polynomial	$y = 1948.08 x - 1366.34 x^{2} - 3.427 x^{3} + 809.243$	0.497	274.047	212.017
	对数Logarithmic	$y = 340.07 l n ? x + 981.64$	0.499	250.078	205.620
	指数Exponential	$y = 596.89456 e x p (0.35384 x)$	0.602	242.616	195.134

Fig.6 Fitting results of optimal model training samples

Fig.7 Scatter plots of optimal model validation samples

References 21

1	Dong S K， Tang F L， Ping X Y， et al. Zoning and functions of China’s grassland in the New Era of ecological civilization. Journal of Natural Resources， 2022， 37（3）： 568-581.
	董世魁，唐芳林，平晓燕，等. 新时代生态文明背景下中国草原分区与功能辨析. 自然资源学报， 2022， 37（3）： 568-581.
2	Piao S L， Fang J Y， He J S， et al. Spatial distribution of grassland biomass in China. Acta Phytoecologica Sinica， 2004， 28（4）： 491-498.
	朴世龙，方精云，贺金生，等. 中国草地植被生物量及其空间分布格局. 植物生态学报， 2004， 28（4）： 491-498.
3	Xu B， Yang X C， Tao W G， et al. Remote sensing monitoring upon the grass production in China. Acta Ecologica Sinica， 2007， 27（2）： 405-413.
	徐斌，杨秀春，陶伟国，等. 中国草原产草量遥感监测. 生态学报， 2007， 27（2）： 405-413.
4	Jin Y X， Xu B， Yang X C， et al. Remote sensing dynamic estimation of grass production in Xilinguole， Inner Mongolia. Scientia Sinica Vitae， 2011， 41（12）： 1185-1195.
	金云翔，徐斌，杨秀春，等. 内蒙古锡林郭勒盟草原产草量动态遥感估算. 中国科学：生命科学， 2011， 41（12）： 1185-1195.
5	Lu Y， Yang S X， Li X H. Monitoring of grassland herbage accumulation by using remote sensing in Gannan Prefecture. Pratacultural Science， 2021， 38（1）： 32-43.
	陆荫，杨淑霞，李晓红. 甘南州高寒天然草地生长状况遥感监测. 草业科学， 2021， 38（1）： 32-43.
6	Huang J X， Wu J， Li C B， et al. Remote sensing retrieval of grassland above-ground biomass in Tianzhu County based on Sentinel-2 and Landsat 8 data. Acta Agrestia Sinica， 2021， 29（9）： 2023-2030.
	黄家兴，吴静，李纯斌，等. 基于Sentinel-2和Landsat 8数据的天祝县草地地上生物量遥感反演. 草地学报， 2021， 29（9）： 2023-2030.
7	Yang D， Yang X C， Jin Y X， et al. Evaluating the research status quo around remote sensing-mediated monitoring of grassland biomass based on bibliometrology. Pratacultural Science， 2021， 38（9）： 1782-1792.
	杨东，杨秀春，金云翔，等. 基于文献计量的草地生物量遥感监测研究进展. 草业科学， 2021， 38（9）： 1782-1792.
8	Garcia M， Riano D， Chuvieco E， et al. Estimating biomass carbon stocks for a Mediterranean forest in central Spain using LiDAR height and intensity data. Remote Sensing of Environment， 2010， 114（4）： 816-830.
9	Wang P H， Xi X H， Wang C， et al. Study on power line fast extraction based on airborne LiDAR data. Science of Surveying and Mapping， 2017， 42（2）： 154-158， 171.
	王平华，习晓环，王成，等. 机载激光雷达数据中电力线的快速提取. 测绘科学， 2017， 42（2）： 154-158， 171.
10	Shen J， Ding L， Xin X P， et al. Canopy scale characteristics of grassland under different grazing intensities based on UAV lidar and multispectral data. Acta Prataculturae Sinica， 2022， 31（3）： 1-15.
	沈洁，丁蕾，辛晓平，等. 基于无人机激光雷达与多光谱数据的不同放牧强度下草原冠层尺度特征研究. 草业学报， 2022， 31（3）： 1-15.
11	Zhang X， Bao Y H， Wang D L， et al. Using UAV LiDAR to extract vegetation parameters of Inner Mongolian grassland. Remote Sensing， 2021， 13（4）： 656.
12	Chen A， Yang X C， Xu B， et al. Research on recognition methods of Elm sparse forest based on object-based image analysis and deep learning. Journal of Geo-information Science， 2020， 22（9）： 1897-1909.
	陈昂，杨秀春，徐斌，等. 基于面向对象与深度学习的榆树疏林识别方法研究. 地球信息科学学报， 2020， 22（9）： 1897-1909.
13	Liu Y， Xi X H， Wang C， et al. An improved progressive TIN densification filtering algorithm for point clouds. Science of Surveying and Mapping， 2020， 45（5）： 106-111， 125.
	刘洋，习晓环，王成，等. 一种改进的渐进加密三角网点云滤波算法. 测绘科学， 2020， 45（5）： 106-111， 125.
14	Zhang X. Research on vegetation biomass inversion based on UAV LiDAR and multispectral data. Hohhot： Inner Mongolia Normal University， 2020.
	张翔. 基于无人机激光雷达和多光谱数据的植被生物量反演研究. 呼和浩特：内蒙古师范大学， 2020.
15	Wang D L， Xin X P， Shao Q Q， et al. Modeling aboveground biomass in Hulunber grassland ecosystem by using unmanned aerial vehicle discrete lidar. Sensors， 2017， 17（1）： 180.
16	Liu X L， Sui L C， Bai Y F， et al. Biomass estimation of Caragana microphylla in the shrub-encroached grassland based on terrestrial laser scanning. Journal of Remote Sensing， 2020， 24（7）： 894-903.
	刘晓亮，隋立春，白永飞，等. 地基激光雷达灌丛化草原小叶锦鸡儿生物量估算. 遥感学报， 2020， 24（7）： 894-903.
17	Chi H， Huang J L， Qiu J， et al. Estimation of forest aboveground biomass using ICESat/GLAS data and Landsat/ETM+ imagery. Science of Surveying and Mapping， 2018， 43（4）： 9-16， 23.
	池泓，黄进良，邱娟，等. GLAS星载激光雷达和Landsat/ETM+数据的森林生物量估算. 测绘科学， 2018， 43（4）： 9-16， 23.
18	Li Z J， Huang B， Lei J G. Analysis of the factors affecting point cloud density of airborne LiDAR. Science of Surveying and Mapping， 2019， 44（6）： 204-211.
	李志杰，黄兵，雷建国. 影响机载激光雷达点云密度的因素分析. 测绘科学， 2019， 44（6）： 204-211.
19	Wang K K， Zheng X D， Lai X D. Relationship between airborne LiDAR point cloud density and DEM product accuracy. Journal of Geomatics， 2021， 46（3）： 78-82.
	王康康，郑学东，赖旭东. 机载LiDAR点云密度与DEM产品精度关系研究. 测绘地理信息， 2021， 46（3）： 78-82.
20	Xie H， Tong X J， Li J， et al. Changes of NDVI and EVI and their responses to climatic variables in the Yellow River Basin during the growing season of 2000-2018. Acta Ecologica Sinica， 2022， 42（11）： 1-14.
	解晗，同小娟，李俊，等. 2000-2018年黄河流域生长季NDVI、EVI变化及其对气候因子的响应. 生态学报， 2022， 42（11）： 1-14.
21	Xing X Y， Yang X C， Xu B， et al. Remote sensing estimation of grassland aboveground biomass based on random forest. Journal of Geo-information Science， 2021， 23（7）： 1312-1324.
	邢晓语，杨秀春，徐斌，等. 基于随机森林算法的草原地上生物量遥感估算方法研究. 地球信息科学学报， 2021， 23（7）： 1312-1324.

[1]	Yi-han ZHAO, Meng-jing HOU, Qi-sheng FENG, Hong-yuan GAO, Tian-gang LIANG, Jin-sheng HE, Da-wen QIAN. Estimation of aboveground biomass in Menyuan grassland based on Landsat 8 and random forest approach [J]. Acta Prataculturae Sinica, 2022, 31(7): 1-14.
[2]	LIANG Tian-gang, CUI Xia, FENG Qi-sheng, WANG Ying, XIA Wen-tao. Remotely sensed dynamics monitoring of grassland aboveground biomass and carrying capacityduring 2001-2008 in Gannan pastoral area [J]. Acta Prataculturae Sinica, 2009, 18(6): 12-22.