Estimating leaf mass per area with statistical and deep learning models

LIU Linlin; YU Ying; YANG Xiguang

LIU Linlin, YU Ying, YANG Xiguang

Journal of Nanjing Forestry University (Natural Sciences Edition） ›› 2025

PDF(2909 KB)

Journal of Nanjing Forestry University (Natural Sciences Edition） ›› 2025

Estimating leaf mass per area with statistical and deep learning models

LIU Linlin, YU Ying^*, YANG Xiguang

Author information +

History +

Abstract

【Objective】Leaf mass per area (LMA) is an important indicator factor reflecting the efficiency of plant resource utilization and environmental response strategies. This study explores model estimation methods for LMA based on different data types and analyzes the spatial variations of LMA under different forest stand types. 【Method】The developed model for LMA estimation was constructed based on the measured leaf spectral reflectance and LMA data of three forest types in Maoershan Forest Farm, Heilongjiang Province, combined with the FASPECT simulation database of the mechanism model, the partial least squares regression (PLSR) model, support vector regression (SVR) model, random forest (RF) model and TabNet model. The influence of different data sources and different models on the simulation accuracy of LMA was compared. Combined with light detection and ranging (LiDAR) point cloud and hyperspectral image data, the distribution characteristics of LMA at stand scale were analyzed. 【Result】The simulation accuracy of the deep learning model TabNet was the best, with an R² of 0.881 and RMSE of 0.001 1 g·cm⁻². The model that combines measured and simulated data still has good stability, with R² and RMSE being 0.987 and 0.001 4 g·cm⁻², respectively. There are significant differences in LMA among different forest types. Coniferous forests have a higher LMA than mixed coniferous and broad-leaved forests and broad-leaved forests. The LMA in the upper and middle layers of the canopy is higher than that in the lower layers. 【Conclusion】The deep learning model TabNet based on the original spectral reflectance data of leaves, which integrates LiDAR point clouds and hyperspectral data, is feasible and applicable for retrieving LMA of different forest stand types in three-dimensional space, providing a reference for improving the accuracy and stability of LMA estimation in large areas.

Key words

hyperspectral / leaf mass per area (LMA) / TabNet model / leaf radiative transfer model / LiDAR

Cite this article

EndNote

Ris (Procite)

Bibtex

Download Citations

LIU Linlin, YU Ying, YANG Xiguang. Estimating leaf mass per area with statistical and deep learning models[J]. Journal of Nanjing Forestry University (Natural Sciences Edition）. 2025

References

[1] 邹晓明, 王国兵, 葛之葳, 等. 林业碳汇提升的主要原理和途径[J]. 南京林业大学学报(自然科学版), 2022, 46(6): 167-176.
ZOU X M, WANG G B, GE Z W, et al.Mechanisms and methods for augmenting carbon sink in forestry[J]. Journal of Nanjing Forestry University (Natural Sciences Edition), 2022, 46(6): 167-176. DOI: 10.12302/j.issn.1000-2006.202209008.
[2] 张林, 罗天祥. 植物叶寿命及其相关叶性状的生态学研究进展[J]. 植物生态学报, 2004, 28(6): 844-852.
ZHANG L, LUO T X.Advances in ecological studies on leaf lifespan and associated leaf traits[J]. Chinese Journal of Plant Ecology, 2004, 28(6): 844-852. DOI: 10.17521/cjpe.2004.0110.
[3] OSNAS J L, LICHSTEIN J W, REICH P B, et al.Global leaf trait relationships: mass, area, and the leaf economics spectrum[J]. Science, 2013, 340(6133), 741-744. DOI: 10.1126/science.1231574.
[4] POORTER L, BONGERS F.Leaf traits are good predictors of plant performance across 53 rain forest species[J]. Ecology, 2006, 87(7), 1733-1743. DOI: 10.1890/0012-9658(2006)87[1733:LTAGPO]2.0.CO;2.
[5] POORTER H, NIINEMETS Ü, POORTER L, et al.Causes and consequences of variation in leaf mass per area (LMA): A meta-analysis[J]. New Phytologist, 2009, 182(3), 565-588. DOI: 10.1111/j.1469-8137.2009.02830.x.
[6] CASTRO-DÍEZ P, PUYRAVAUD J P, CORNELISSEN J H C. Leaf structure and anatomy as related to leaf mass per area variation in seedlings of a wide range of woody plant species and types[J]. Oecologia, 2000, 124(4): 476-486. DOI: 10.1007/PL00008873.
[7] JI F J, LI F, DASHTI H, et al.Leveraging transfer learning and leaf spectroscopy for leaf trait prediction with broad spatial, species, and temporal applicability[J]. Remote Sensing of Environment, 2025, 326: 114818-114818. DOI: 10.1016/J.RSE.2025.114818.
[8] 刘明秀, 梁国鲁. 植物比叶质量研究进展[J]. 植物生态学报, 2016, 40(8): 847-860.
LIU M X, LIANG G L.Research progress on leaf mass per area[J]. Chinese Journal of Plant Ecology, 2016, 40(8): 847-860. DOI: 10.17521/cjpe.2015.0428.
[9] PAJARES G.Overview and current status of remote sensing applications based on Unmanned aerial vehicles (UAVs) photogramm[J]. Photogrammetric Engineering & Remote Sensing, 2015, 81(4): 281-330. DOI: 10.14358/PERS.81.4.281.
[10] GROSSKINSKY D K, SVENSGAARD J, CHRISTENSEN S, et al.Plant phenomics and the need for physiological phenotyping across scales to narrow the genotype-to-phenotype knowledge gap[J]. Journal of Experimental Botany, 2015, 66(18): 5429-5440. DOI: 10.1093/jxb/erv345.
[11] 赵国帅. 无人机遥感在林业中的应用与需求分析[J]. 福建林业科技, 2017, 44(1): 136-140.
ZHAO G S.Application and demand analysis of unmanned aerial vehicle remote sensing in forestry[J]. Journal of Fujian Forestry Science and Technology, 2017, 44(1): 136-140. DOI: 10.13428/j.cnki.fjlk.2017.01.029.
[12] SHEN X, CAO L, COOPS N C, et al.Quantifying vertical profiles of biochemical traits for forest plantation speciesusing advanced remote sensing approaches[J]. Remote Sensing of Environment, 2020, 250:112041. DOI: 10.1016/j.rse.2020.112041.
[13] 潘磊, 孙玉军, 王轶夫, 等. 基于Sentinel-1和Sentinel-2数据的杉木林地上生物量估算[J]. 南京林业大学学报(自然科学版), 2020, 44(3): 149-156.
PAN L, SUN Y J, WANG Y F, et al.Estimation of aboveground biomass in a Chinese fir (Cunning-hamia lanceolata) forest combining data of Sentinel-1 and Sentinel-2[J]. Journal of Nanjing Forestry University (Natural Sciences Edition), 2020, 44(3): 149-156. DOI: 10.3969/j.issn.1000-2006.201811012.
[14] 杨涛, 于颖, 杨曦光. 无人机高光谱联合LiDAR估测林分与单木尺度叶绿素含量[J]. 应用生态学报, 2023, 34(8): 2101-2112.
YANG T, YU Y, YANG X G.UAV hyperspectral combined with LiDAR to estimate chlorophyll content at the stand and single tree scale[J]. Chinese Journal of Applied Ecology, 2023, 34(8): 2101-2112. DOI: 10.13287/j.1001-9332.202308.004.
[15] JUOLA J, HOVI A, RAUTIAINEN M.Comparison of contemporaneous Sentinel-2 and EnMAP data for vegetation index-based estimation of leaf area index and canopy closure of a boreal forest[J]. European journal of remote sensing, 2024, 57(1): 2432975. DOI: 10.1080/22797254.2024.2432975.
[16] 包广道, 刘婷, 张忠辉, 等. 长白山区4种针叶林有效叶面积指数遥感精细反演及空间分布规律[J]. 林业科学, 2024, 60(5): 127-138.
BAO G D, LIU T, ZHANG Z H, et al.Remote sensing inversion of effective leaf area index of four coniferous forest types and their spatial distribution rule in changbai mountain[J]. Scientia Silvae Sinicae, 2024, 60(5):127-138. DOI: 10.11707/j.1001-7488.LYKX20220545.
[17] 孙嘉. 高光谱激光雷达植被生化参数遥感定量反演[D]. 武汉大学, 2019.
SUN J.Quantitive remote sensing of vegetation biochemical parameters by hyperspectral LiDAR[D]. Wuhan University, 2019. DOI: 10.27379/d.cnki.gwhdu.2019.001014.
[18] SINGH A, SERBIN S P, MCNEIL B E, et al.Imaging spectroscopy algorithms for mapping canopy foliar chemical and morphological traits and their uncertainties[J]. Ecological applications, 2015, 25(8): 2180-2197. DOI: 10.1890/14-2098.1.
[19] SERBIN S P, WU J, Ely K S, et al.From the Arctic to the tropics: multibiome prediction of leaf mass per area using leaf reflectance[J]. The New phytologist, 2019, 224(4): 1557-1568. DOI: 10.1111/nph.16123.
[20] 刘霜. 基于Sentinel-1/2的重庆市南川区森林生物量估算研究[D]. 成都理工大学, 2020.
LIU S.Forest biomass estimation in nanchuan district of Chongqing city using a combination of Sentinel-1 and Sentinel-2 data[D]. Chengdu University of Technology, 2020. DOI: 10.26986/d.cnki.gcdlc.2020.000457.
[21] 史博太, 常庆瑞, 崔小涛, 等. 基于Sentinel-2多光谱数据和机器学习算法的冬小麦LAI遥感估算[J]. 麦类作物学报, 2021, 41(6): 752-761.
SHI B T, CHANG Q R, CUI X T, et al.LAI estimation of winter wheat based on Sentinel-2 multis-pectral data and machine learning algorithm[J]. Journal of Triticeae Crops, 2021, 41(6): 752-761. DOI: 10.7606/j.issn.1009-1041.2021.06.14.
[22] 刘涛, 张寰, 王志业, 等. 利用无人机多光谱估算小麦叶面积指数和叶绿素含量[J]. 农业工程学报, 2021, 37(19): 65-72.
LIU T, ZHANG H, WANG Z Y, et al.Estimation of the leaf area index and chlorophyll content of wheat using UAV multi-spectrum images[J]. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(19): 65-72. DOI: 10.11975/j.issn.1002-6819.2021.19.008.
[23] 陈小平, 王树东, 张立福, 等. 植被叶片含水量反演的精度及敏感性[J]. 遥感信息, 2016, 31(1): 48-57.
CHEN X P, WANG S D, ZHANG L F, et al.Accuracy and sensitivity of retrieving vegetation leaf water content[J]. Remote Sensing Information, 2016, 31(1): 48-57. DOI: 10.3969/j.issn.1000-3177.2016.01.008.
[24] 王洋, 肖文, 邹焕成, 等. 基于PROSPECT模型的植物叶片干物质估测建模研究[J]. 沈阳农业大学学报, 2018, 49(01): 121-127.
WANG Y, XIAO W, ZOU H C, et al.Plant leaf dry matter estimating and modeling based on PROSPECTmodel[J]. Journal of Shenyang Agricultural University, 2018, 49(01): 121-127. DOI: 10.3969/j.issn.1000-1700.2018.01.018.
[25] ARIK S Ö, PFISTER T.TabNet: Attentive Interpretable Tabular Learning[J]. Proceedings of the AAAI ConferenceonArtificial Intelligence, 2021, 35(8): 6679-6687. DOI: https://doi.org/10.1609/aaai.v35i8.16826.
[26] JACQUEMOUD S, BARET F.PROSPECT: A model of leaf optical properties spectra[J]. Remote Sensing of Environment, 1990, 34(2): 75-91. DOI: 10.1016/0034-4257(90)90100-Z.
[27] QIU F, CHEN J M, JU W, et al.Improving the PROSPECT model to consider anisotropic scattering of leaf internalmaterials and its use for retrieving leaf biomass in fresh leaves[J]. IEEE Transactions on geoscience and remote sensing, 2018, 56(6): 3119-3136. DOI: 10.1109/tgrs.2018.2791930.
[28] JIANG J Y, COMAR A, WEISS M, et al.FASPECT: A model of leaf optical properties accounting for the differences between upper and lower faces[J]. Remote Sensing of Environment, 2021, 253: 112-205. DOI: 10.1016/j.rse.2020.112205.
[29] SHI H Y, JIANG J Y, JACQUEMOUD S, et al.Estimating leaf mass per area with leaf radiative transfer model[J]. Remote Sensing of Environment, 2023, 286:113444. DOI: 10.1016/j.rse.2022.113444.
[30] SUN J, YANG J, SHI S, et al.Estimating rice leaf nitrogen concentration: influence of regression algorithms based on passive and active leaf reflectance[J]. Remote Sensing, 2017, 9(9): 951-951. DOI: 10.3390/rs9090951.
[31] LU B, He Y H.Evaluating empirical regression, machine learning, and radiative transfer modelling for estimating vegetation chlorophyll content using bi-seasonal hyperspectral images[J]. Remote Sensing, 2019, 11(17): 1979-1979. DOI: 10.3390/rs11171979.
[32] LUO S Z, CHEN J M, WANG C, et al.Comparative performances of airborne LiDAR height and intensity data for leaf area index estimation[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2017, 11(1): 300-310. DOI: 10.1109/jstars.2017.2765890.
[33] YIN S Y, ZHOU K, CAO L, et al.Estimating the horizontal and vertical distributions of pigments in canopies of ginkgo plantation based on UAV-borne LiDAR, hyperspectral data by coupling PROSAIL model[J]. Remote Sensing, 2022, 14(3): 715-715. DOI: 10.3390/RS14030715.
[34] LICHTENTHALER H K, BABANI F, LANGSDORF G.Chlorophyll fluorescence imaging of photosynthetic activity in sun and shade leaves of trees[J]. Photosynthesis Research, 2007, 93(1-3), 235-244. DOI: 10.1007/s11120-007-9174-0.
[35] IMPOLLONIA G, CROCI M, BLANDINIÈRES H, et al. Comparison of PROSAIL model inversion methods for estimating leaf chlorophyll content and LAI using UAV imagery for hemp phenotyping[J]. Remote Sensing, 2022, 14(22): 5801-5801. DOI: 10.3390/RS14225801.

PDF(2909 KB)

Accesses

Citation

Detail

Sections

Recommended

The full text is translated into English by AI, aiming to facilitate reading and comprehension. The core content is subject to the explanation in Chinese.

www.nldxb.njfu.edu.cn

Edited ＆ Published by Editorial Department of Nanjing Forestry University(Natural Sciences Edition)
Address： No.159 Longpan Road,Nanjing 210037 Jiangsu,P.R.China
E-mail ：xuebao@njfu.edu.cn , xuebao@njfu.com.cn
Telephone +86-25-85428247,85427076
Distributed by China International Book Trading Corporation P.O. Box 399, Beijing ,P.R. China

〈

〉

Please choose a citation manager

Content to export