Estimation of aboveground biomass using in situ hyperspectral measurements in five major grassland ecosystems on the Tibetan Plateau

doi:10.1093/jpe/rtn025

Abstract

Abstract: Aims There are numerous grassland ecosystem types on the Tibetan Plateau. These include the alpine meadow and steppe and degraded alpine meadow and steppe. This study aimed at developing a method to estimate aboveground biomass (AGB) for these grasslands from hyperspectral data and to explore the feasibility of applying air/satellite-borne remote sensing techniques to AGB estimation at larger scales.
Methods We carried out a field survey to collect hyperspectral reflectance and AGB for five major grassland ecosystems on the Tibetan Plateau and calculated seven narrow-band vegetation indices and the vegetation index based on universal pattern decomposition (VIUPD) from the spectra to estimate AGB. First, we investigated correlations between AGB and each of these vegetation indices to identify the best estimator of AGB for each ecosystem type. Next, we estimated AGB for the five pooled ecosystem types by developing models containing dummy variables. At last, we compared the predictions of simple regression models and the models containing dummy variables to seek an ecosystem type-independent model to improve prediction of AGB for these various grassland ecosystems from hyperspectral measurements.
Important findings When we considered each ecosystem type separately, all eight vegetation indices provided good estimates of AGB, with the best predictor of AGB varying among different ecosystems. When AGB of all the five ecosystems was estimated together using a simple linear model, VIUPD showed the lowest prediction error among the eight vegetation indices. The regression models containing dummy variables predicted AGB with higher accuracy than the simple models, which could be attributed to the dummy variables accounting for the effects of ecosystem type on the relationship between AGB and vegetation index (VI). These results suggest that VIUPD is the best predictor of AGB among simple regression models. Moreover, both VIUPD and the soil-adjusted VI could provide accurate estimates of AGB with dummy variables integrated in regression models. Therefore, ground-based hyperspectral measurements are useful for estimating AGB, which indicates the potential of applying satellite/airborne remote sensing techniques to AGB estimation of these grasslands on the Tibetan Plateau.

Key words: biomass estimation, dummy variable, hyperspectral remote sensing, Tibetan Plateau, regression analysis, vegetation index, VIUPD

摘要：
Aims There are numerous grassland ecosystem types on the Tibetan Plateau. These include the alpine meadow and steppe and degraded alpine meadow and steppe. This study aimed at developing a method to estimate aboveground biomass (AGB) for these grasslands from hyperspectral data and to explore the feasibility of applying air/satellite-borne remote sensing techniques to AGB estimation at larger scales.
Methods We carried out a field survey to collect hyperspectral reflectance and AGB for five major grassland ecosystems on the Tibetan Plateau and calculated seven narrow-band vegetation indices and the vegetation index based on universal pattern decomposition (VIUPD) from the spectra to estimate AGB. First, we investigated correlations between AGB and each of these vegetation indices to identify the best estimator of AGB for each ecosystem type. Next, we estimated AGB for the five pooled ecosystem types by developing models containing dummy variables. At last, we compared the predictions of simple regression models and the models containing dummy variables to seek an ecosystem type-independent model to improve prediction of AGB for these various grassland ecosystems from hyperspectral measurements.
Important findings When we considered each ecosystem type separately, all eight vegetation indices provided good estimates of AGB, with the best predictor of AGB varying among different ecosystems. When AGB of all the five ecosystems was estimated together using a simple linear model, VIUPD showed the lowest prediction error among the eight vegetation indices. The regression models containing dummy variables predicted AGB with higher accuracy than the simple models, which could be attributed to the dummy variables accounting for the effects of ecosystem type on the relationship between AGB and vegetation index (VI). These results suggest that VIUPD is the best predictor of AGB among simple regression models. Moreover, both VIUPD and the soil-adjusted VI could provide accurate estimates of AGB with dummy variables integrated in regression models. Therefore, ground-based hyperspectral measurements are useful for estimating AGB, which indicates the potential of applying satellite/airborne remote sensing techniques to AGB estimation of these grasslands on the Tibetan Plateau.

Miaogen Shen, Yanhong Tang, Julia Klein, Pengcheng Zhang, Song Gu, Ayako Shimono, Jin Chen. Estimation of aboveground biomass using in situ hyperspectral measurements in five major grassland ecosystems on the Tibetan Plateau[J]. J Plant Ecol, 2008, 1(4): 247-257.

[1]	Yan Zhang, Yanhong Tang. Flower surface is warmer in center than at edges in alpine plants: evidence from Qinghai-Tibetan Plateau [J]. J Plant Ecol, 2023, 16(6): 0-rtad023.
[2]	Shuping Guan, Pengdong Chen, Xingle Qu, Xiaolan Wang, Shuopeng Wang, Haiying Li, Jiangping Fang, Yi Wang, Jiarui Chen, Wei Huang, Evan Siemann. Invasiveness and impact of invasive species on the Tibetan Plateau are inconsistent [J]. J Plant Ecol, 2023, 16(6): 0-rtad039.
[3]	Le Sun, Jian Sun, Jun-Xi Wu, Zi-Yin Du, You-Jun Chen, Yi Wang, Miao Liu, Wen-Cheng Li, Er-Yuan Liang. Plant community traits and functions mediate the biomass trade-off of alpine grasslands along precipitation gradients on the Tibetan Plateau [J]. J Plant Ecol, 2023, 16(5): 0-rtad009.
[4]	Yang Lin, Jiang-Tao Xiao, Yong-Ping Kou, Jia-Xing Zu, Xin-Ran Yu, Yuan-Yuan Li. Aboveground carbon sequestration rate in alpine forests on the eastern Tibetan Plateau: impacts of future forest management options [J]. J Plant Ecol, 2023, 16(3): 0-rtad001.
[5]	Li Mo, Hao Yang, Peng Luo, Cheng-Xiang Mou, Jun Wang. The joint effects of resource stress and grazing on plant-plant interactions in alpine meadows on the eastern Qinghai-Tibetan Plateau [J]. J Plant Ecol, 2023, 16(2): 0-rtac062.
[6]	Bowen Li, Jianping Sun, Shiping Wang, Wangwang Lv, Yang Zhou, Peipei Liu, Qi Wang, Wang A, Suren Zhang, Lu Xia, Huan Hong, Lili Jiang, Caiyun Luo, Zhenhua Zhang, Shilong Piao, Yanfen Wang, Tsechoe Dorji. Greater responses of flower phenology of Kobresia pygmaea community to precipitation addition than to constant and stepwise warming [J]. J Plant Ecol, 2023, 16(2): 0-rtac066.
[7]	Yang Zhou, Wang-Wang Lv, Shi-Ping Wang, Li-Rong Zhang, Jian-Ping Sun, Li-Li Jiang, Pei-Pei Liu, Qi Wang, Bo-Wen Li, A. Wang, Huan Hong, Su-Ren Zhang, Lu Xia, Nan Ji, Zheng-Xin Xie, Cai-Yun Luo, Zhen-Hua Zhang, Chang-Hui Wang, Jin-Zhi Wang, Ci Yang and Tsechoe Dorji. Additive effects of warming and grazing on fine-root decomposition and loss of nutrients in an alpine meadow [J]. J Plant Ecol, 2022, 15(6): 1273-1284.
[8]	Tao Zhang, Ximeng Ji, Yuanyuan Tang, Mingjie Xu, Yangjian Zhang, Guang Zhao, Ning Chen, Juntao Zhu and Yongtao He. Fisher discriminant analysis method applied in drought detection: an instance in an alpine meadow ecosystem [J]. J Plant Ecol, 2022, 15(5): 911-920.
[9]	Mingjie Xu, Tingting An, Zhoutao Zheng, Tao Zhang, Yangjian Zhang and Guirui Yu. Variability in evapotranspiration shifts from meteorological to biological control under wet versus drought conditions in an alpine meadow [J]. J Plant Ecol, 2022, 15(5): 921-932.
[10]	Mei Liu, Jia-Hao Wen, Ya-Mei Chen, Wen-Juan Xu, Qiong Wang and Zhi-Liang Ma. Warming increases soil carbon input in a Sibiraea angustata-dominated alpine shrub ecosystem [J]. J Plant Ecol, 2022, 15(2): 335-346.
[11]	Xin Chen, Qi Li, Dongdong Chen, Fuquan He, Lili Huo, Liang Zhao and Chunwang Xiao. Reasonable management of perennial planting grassland contributes to positive succession of soil microbial community in Sanjiangyuan of Qinghai-Tibetan Plateau, China [J]. J Plant Ecol, 2022, 15(2): 359-371.
[12]	Zhi Zheng, Yue Zhang, Shihu Zhang, Qun Ma, Dajie Gong and Guoying Zhou. Above- and belowground trait linkages of dominant species shape responses of alpine steppe composition to precipitation changes in the Tibetan Plateau [J]. J Plant Ecol, 2021, 14(4): 569-579.
[13]	Ming-Hua Song, Johannes H.C. Cornelissen, Yi-Kang Li, Xing-Liang Xu, Hua-Kun Zhou, Xiao-Yong Cui, Yan-Fen Wang, Rong-Yan Xu and Qi Feng. Small-scale switch in cover–perimeter relationships of patches indicates shift of dominant species during grassland degradation [J]. J Plant Ecol, 2020, 13(6): 704-712.
[14]	Rémi Perronne, Franck Jabot and Julien Pottier. Inter- and intraspecific variability of plant individual growth and its role on species ranking in grasslands [J]. J Plant Ecol, 2020, 13(3): 378-386.
[15]	Aamir Latif, Sana Ilyas, Yangjian Zhang, Yuqin Xin, Lin Zhou and Quan Zhou. Review on global change status and its impacts on the Tibetan Plateau environment [J]. J Plant Ecol, 2019, 12(6): 917-930.