Estimation of Potato Above Ground Biomass Based on UAV Multispectral Images
LIU Yang1, 2, 4, SUN Qian1, 4, HUANG Jue2, FENG Hai-kuan1, 3, 4*, WANG Jiao-jiao1, 4, YANG Gui-jun1, 4
1. Key Laboratory of Quantitative Remote Sensing in Agriculture, Ministry of Agriculture, Beijing Research Center for Information Technology in Agriculture, Beijing 100097, China
2. College of Surveying Science and Engineering, Shandong University of Science and Technology, Qingdao 266590, China
3. National Information Agriculture Engineering Technology Center of Nanjing Agricultural University, Nanjing 210095, China
4. National Engineering Research Center for Information Technology in Agriculture, Beijing 100097, China
摘要: 地上生物量(AGB)是评估作物生长发育和指导田间农业生产管理的重要指标。因此,高效精准地获取作物AGB信息,可以及时准确地估算产量,对于保障粮食供应和贸易提供有力依据。传统获取AGB的方法是采用破坏性取样法,这使得大面积、长期的测量变为困难。然而,随着精准农业的快速发展,无人机遥感技术被认为是估算大面积作物AGB最有效的技术方式。通过无人机平台搭载多光谱传感器获取马铃薯块茎形成期、块茎增长期和淀粉积累期的多光谱影像,地面实测株高和AGB以及地面控制点(GCP)的空间位置信息。首先,基于SFM(structure from motion,SFM)技术利用无人机影像数据结合GCP的三维坐标生成试验田的DSM(digital surface model,DSM),通过DSM提取出马铃薯各生育期的株高(Hdsm);然后,选取原始4个单波段植被指数、9个多波段组合的植被指数、红边波段的高频信息(HFI)和提取的Hdsm分别与AGB作相关性分析;最后基于单波段植被指数(x1)、多波段组合的植被指数(x2)、植被指数结合Hdsm(x3)、植被指数结合HFI(x4)以及植被指数融合HFI和Hdsm(x5)为模型输入参数,采用偏最小二乘回归(PLSR)和岭回归(RR)估算各生育期的AGB。结果表明:(1)提取的Hdsm和实测株高拟合的R2为0.87,NRMSE为14.34%;(2)各模型参数都与AGB达到极显著水平,相关性均从块茎形成期到淀粉积累期先升高后降低;(3)各生育期以5种变量使用同种方法估算马铃薯AGB的效果,均从块茎形成期到淀粉积累期先好后变差,其估算精度由高到低依次为x5>x4>x3>x2>x1;(4)各生育期使用PLSR以不同变量估算AGB的效果要优于RR方法,其中在块茎增长期基于x5变量估算马铃薯AGB效果最佳,R2为0.73,NRMSE为15.22%。因此,选取多光谱植被指数结合红边波段的高频信息和Hdsm并使用PLSR方法可以明显提高AGB的估算精度,这为大面积马铃薯作物AGB的监测提供了新的技术支撑。
关键词:马铃薯;多光谱;株高;植被指数;高频信息;地上生物量
Abstract:Above ground biomass (AGB) is an important indicator of evaluating crop growth and guiding agricultural production and management. Therefore, AGB information was obtained timely, accurately and efficiently to provide a strong basis for predicting yields and securing grain trade. The conventional way to obtain AGB is to use destructive sampling methods that require manual harvesting of crops, weighing, and recording, making large-area and long-term measurements difficult. However, UAV remote sensing technology is considered the most effective way to estimate AGB of large area crops with the rapid development of precision agriculture. In this study, the multispectral images of the tuber formation period, tuber growth period and starch accumulation period were obtained by the UAV platform equipped with multispectral sensors. The measured plant height, AGB and latitude, longitude and altitude of ground control point (GCP) were measured on the ground. Firstly, using UAV multispectral images combined GCP location information basing structure from motion (SFM) algorithm to generate the digital surface model (DSM) of the potato experimental field, and DSM extracted the plant height (Hdsm) of each growth period. Then, four original single band vegetation indices, 9 multiband vegetation indices,high-frequency information (HFI) in the red edge band and Hdsm were selected with AGB for correlation analysis. Finally, based on single-band vegetation indices (x1), multiband vegetation indices (x2), vegetation indicescombined Hdsm (x3),vegetation indices combined HFI (x4) and their integration (x5) as input parameters were used to estimate AGB of each growth period by partial least squares regression (PLSR) and ridge regression (RR). The results showed that: (1) The R2 of extracted Hdsm and measured plant height was 0.87 and NRMSE was 14.34%. (2) All model parameters reached highly significant levels with the AGB, and correlations increased and then decreased from the tuber formation period to the starch accumulation period. (3) Using the same method to estimate potato AGB with five variables at different growth periods, it starts to get better and then it gets worse for the effect of potato AGB from tuber formation period to starch accumulation period with the estimation accuracy from high to low was x5>x4>x3>x2>x1. (4) The results showed that PLSR was better than RR in estimating AGB for different growth stages and basing x5 combined PLSR method was the best in estimating AGB at tuber growth period with R2 of 0.73 and NRMSE of 15.22%. Therefore, this study combined the selected multispectral vegetation indices combined HFI and Hdsm with the PLSR method can significantly improve the estimation accuracy of AGB, which provides new technical support for the monitoring of AGB in large areas of potato crops.
Key words:Potato; Multispectral; Plant height; Vegetation indices; High frequency information; Above ground biomass
刘 杨,孙 乾,黄 珏,冯海宽,王娇娇,杨贵军. 无人机多光谱影像的马铃薯地上生物量估算[J]. 光谱学与光谱分析, 2021, 41(08): 2549-2555.
LIU Yang, SUN Qian, HUANG Jue, FENG Hai-kuan, WANG Jiao-jiao, YANG Gui-jun. Estimation of Potato Above Ground Biomass Based on UAV Multispectral Images. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2021, 41(08): 2549-2555.
[1] Li B, Xu X, Zhang L, et al. ISPRS Journal of Photogrammetry and Remote Sensing, 2020, 162(4): 161.
[2] ZHOU Min-gu, SHAO Guo-min, ZHANG Li-yuan, et al(周敏姑, 邵国敏, 张立元, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2020, 36(20): 125.
[3] YANG Bao-hua, GAO Yuan, WANG Meng-xuan, et al(杨宝华, 高 远, 王梦玄,等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2021, 41(3): 936.
[4] Han L, Yang G, Dai H, et al. Plant Methods, 2019, 15(1): 10.
[5] Bendig J, Kang Y, Aasen H, et al. International Journal of Applied Earth Observation and Geoinformation, 2015, 39(1): 79.
[6] Tao H L, Feng H K, Xu L J, et al. Sensors, 2020, 20(4): 1231.
[7] Qi H X, Zhu B Y, Wu Z Y, et al. Sensors, 2020, 20(23): 6732.
[8] XIAO Wu, CHEN Jia-le, DAN Hong-zhi, et al(肖 武, 陈佳乐, 笪宏志, 等). Transactions of the Chinese Society for Agricultural Machinery(农业机械学报), 2018, 49(8): 169.
[9] SUN Shi-rui, ZHAO Yan-ling, WANG Ya-juan, et al(孙诗睿, 赵艳玲, 王亚娟, 等). Journal of China Agricultural University(中国农业大学学报), 2019, 24(11): 51.
[10] Zheng H B, Li W, Jiang J L, et al. Remote Sensing, 2018, 10(12): 2026.
[11] Brinkhoff J, Dunn B W, Robson A J, et al. Remote Sensing, 2019, 11(15): 1837.
[12] CHEN Peng, FENG Hai-kuan, LI Chang-chun, et al(陈 鹏, 冯海宽, 李长春, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2019, 35(11): 63.
[13] XI Xue, ZHAO Geng-xing(奚 雪, 赵庚星). Chinese Agricultural Science Bulletin(中国农学通报), 2020, 36(20): 119.
[14] LIU Yang, FENG Hai-kuan, HUANG Jue, et al(刘 杨, 冯海宽, 黄 珏, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2021, 41(3): 903.
[15] Bispo P C, Pedro R V, Zimbres B et al. Remote Sensing, 2020, 41(17): 2685.
[16] Masjedi A, Crawford M, Carpenter N R, et al. Remote Sensing, 2020, 12(21): 3587.
[17] Tao H L, Feng H K, Xu J L, et al. Sensors, 2020, 20(5): 1296.
[18] Vuorinne I, Heiskanen J, Pellikka P E. Remote Sensing, 2021, 13(2): 233.
[19] Yue J B, Feng H K, Jin X L, et al. Remote Sensing, 2018, 10(7): 1138.