Explainable machine learning for tracking spatial variation in leaf chlorophyll fluorescence within temperate deciduous forest canopies

Abstract

Monitoring forest ecosystems is essential for understanding ecological processes, assessing biodiversity, and guiding sustainable forest management. Chlorophyll fluorescence (ChlF) parameters provide valuable, physiologically relevant insights into the state of the photosynthetic apparatus and serve as indicators in forest monitoring. However, effectively describing the spatial distribution of ChlF parameters within the canopy and developing convenient methods to track their spatial variations remains challenging for scale extension and remote sensing applications. This study focused on several typical temperate deciduous forest species, for which the ChlF parameters, spectral reflectance, and canopy leaf coordinates were synchronously recorded. The results indicated that there were significant spatial variations in ChlF parameters within the canopy. The development of different machine learning (ML) algorithms for estimating ChlF parameters using different combinations of spectral reflectance, species, and leaf coordinates demonstrated that the random forest and extreme gradient boosting (XGBoost) models using leaf spectral reflectance have considerable potential for capturing spatial variation in ChlF parameters. Furthermore, the inclusion of additional information, such as species and leaf coordinates, significantly improved the performance of these models (explaining over 85% of the variance for all ChlF parameters) as well. This study supports the assessment of spatial patterns of ChlF parameters and demonstrates the potential of ML to characterize their variation across forest canopies. By directly modeling ChlF parameters at high spatial resolution, this approach provides new insights into forest physiological processes and advances biodiversity and environmental research.

Keywords

Canopy / Chlorophyll fluorescence / Machine learning / Spatial variation / Spectral reflectance

References

[1] Allaire JJ, Chollet F (2018) Keras: R interface to ‘Keras’. R package version 2.6. 1. https://CRAN.R-project.org/package=keras

[2] Azadnia R, Rajabipour A, Jamshidi B, Omid M. New approach for rapid estimation of leaf nitrogen, phosphorus, and potassium contents in apple-trees using Vis/NIR spectroscopy based on wavelength selection coupled with machine learning. Comput Electron Agric, 2023, 207: 107746

[3] Baker NR. Chlorophyll fluorescence: a probe of photosynthesis in vivo. Annu Rev Plant Biol, 2008, 59: 89-113

[4] Bartold M, Kluczek M. A machine learning approach for mapping chlorophyll fluorescence at inland wetlands. Remote Sens, 2023, 15(9): 2392

[5] Bartold M, Kluczek M. Estimating of chlorophyll fluorescence parameter Fv/Fm for plant stress detection at peatlands under Ramsar Convention with Sentinel-2 satellite imagery. Ecol Inform, 2024, 81: 102603

[6] Bartold M, Wróblewski K, Kluczek M, Dąbrowska-Zielińska K, Goliński P. Examining the sensitivity of satellite-derived vegetation indices to plant drought stress in grasslands in Poland. Plants Basel, 2024, 13(16): 2319

[7] Bilger W, Björkman O. Role of the xanthophyll cycle in photoprotection elucidated by measurements of light-induced absorbance changes, fluorescence and photosynthesis in leaves of Hedera canariensis. Photosynth Res, 1990, 25(3): 173-185

[8] Bischl B, Richter J, Bossek J, Horn D, Thomas J, Lang M (2017) mlrMBO: a modular framework for model-based optimization of expensive black-box functions. arXiv: 1703.03373. https://arxiv.org/abs/1703.03373

[9] Bochenek Z, Ziolkowski D, Bartold M, Orlowska K, Ochtyra A. Monitoring forest biodiversity and the impact of climate on forest environment using high-resolution satellite images. Eur J Remote Sens, 2018, 51(1): 166-181

[10] Bochenek Z, Ziolkowski D, Bartold M (2015) Forest condition assessment through analyzing relations between meteorological parameters describing climate changes and vegetation indices derived from low-resolution satellite data. In: 2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS). Milan, Italy. pp 3341–3344. https://doi.org/10.1109/IGARSS.2015.7326534

[11] Čepl J, Holá D, Stejskal J, Korecký J, Kočová M, Lhotáková Z, Tomášková I, Palovská M, Rothová O, Whetten RW, Kaňák J, Albrechtová J, Lstibůrek M. Genetic variability and heritability of chlorophyll a fluorescence parameters in Scots pine. Tree Physiol, 2016, 36(7): 883-895

[12] Chang CY, Wen JM, Han JM, Kira O, LeVonne J, Melkonian J, Riha SJ, Skovira J, Ng S, Gu LH, Wood JD, Näthe P, Sun Y. Unpacking the drivers of diurnal dynamics of sun-induced chlorophyll fluorescence (SIF): canopy structure, plant physiology, instrument configuration and retrieval methods. Remote Sens Environ, 2021, 265: 112672

[13] Chen T, He T, Benesty M, Khotilovich V, Tang Y, Cho H, Chen K, Mitchell R, Cano I, Zhou T (2015) Xgboost: extreme gradient boosting. R package version 0.4–2

[14] Deepak M, Keski-Saari S, Fauch L, Granlund L, Oksanen E, Keinänen M. Leaf canopy layers affect spectral reflectance in silver birch. Remote Sens, 2019, 11(24): 2884

[15] Escudero A, Fernández J, Cordero A, Mediavilla S. Distribution of leaf characteristics in relation to orientation within the canopy of woody species. Acta Oecol, 2013, 48: 13-20

[16] Féret JB, le Maire G, Jay S, Berveiller D, Bendoula R, Hmimina G, Cheraiet A, Oliveira JC, Ponzoni FJ, Solanki T, de Boissieu F, Chave J, Nouvellon Y, Porcar-Castell A, Proisy C, Soudani K, Gastellu-Etchegorry JP, Lefèvre-Fonollosa MJ. Estimating leaf mass per area and equivalent water thickness based on leaf optical properties: potential and limitations of physical modeling and machine learning. Remote Sens Environ, 2019, 231: 110959

[17] Ferguson JN, Jithesh T, Lawson T, Kromdijk J. Excised leaves show limited and species-specific effects on photosynthetic parameters across crop functional types. J Exp Bot, 2023, 74(21): 6662-6676

[18] Foley S, Rivard B, Sanchezazofeifa G, Calvo J. Foliar spectral properties following leaf clipping and implications for handling techniques. Remote Sens Environ, 2006, 103(3): 265-275

[19] Fotis AT, Curtis PS. Effects of structural complexity on within-canopy light environments and leaf traits in a northern mixed deciduous forest. Tree Physiol, 2017, 37(10): 1426-1435

[20] Friedman J, Hastie T, Tibshirani R. Regularization paths for generalized linear models via coordinate descent. J Stat Softw, 2010, 33(1): 1-22

[21] Gara TW, Darvishzadeh R, Skidmore AK, Wang TJ. Impact of vertical canopy position on leaf spectral properties and traits across multiple species. Remote Sens, 2018, 10(2): 346

[22] Gaur S, Drewry DT. Explainable machine learning for predicting stomatal conductance across multiple plant functional types. Agric for Meteorol, 2024, 350: 109955

[23] Gitelson AA, Buschmann C, Lichtenthaler HK. The chlorophyll fluorescence ratio F₇₃₅/F₇₀₀ as an accurate measure of the chlorophyll content in plants. Remote Sens Environ, 1999, 69(3): 296-302

[24] Gu LH, Han JM, Wood JD, Chang CY, Sun Y. Sun-induced Chl fluorescence and its importance for biophysical modeling of photosynthesis based on light reactions. New Phytol, 2019, 223(3): 1179-1191

[25] Gurdak R, Bartold M. Remote sensing techniques to assess chlorophyll fluorescence in support of crop monitoring in Poland. Misc Geogr, 2021, 25(4): 226-237

[26] Hallik L, Niinemets KO. Photosynthetic acclimation to light in woody and herbaceous species: a comparison of leaf structure, pigment content and chlorophyll fluorescence characteristics measured in the field. Plant Biol, 2012, 14(1): 88-99

[27] Huang XJ, Zheng Y, Zhang H, Lin SR, Liang SL, Li XQ, Ma MG, Yuan WP. High spatial resolution vegetation gross primary production product: algorithm and validation. Sci Remote Sens, 2022, 5: 100049

[28] Iio A, Fukasawa H, Nose Y, Kato S, Kakubari Y. Vertical, horizontal and azimuthal variations in leaf photosynthetic characteristics within a Fagus crenata crown in relation to light acclimation. Tree Physiol, 2005, 25(5): 533-544

[29] Ishida A, Toma T, Marjenah (1999) Leaf gas exchange and cholorphyll fluorescence in relation to leaf angle, azimuth, and canopy position in the tropical pioneer tree, Macaranga conifer, Tree Physiol 19(2): 117–124. https://doi.org/10.1093/treephys/19.2.117

[30] Jin J, Wang Q. Hyperspectral indices developed from the low order fractional derivative spectra can capture leaf dry matter content across a variety of species better. Agric for Meteorol, 2022, 322: 109007

[31] Jin J, Pratama BA, Wang Q, Jin J, Pratama BA, Wang Q. Tracing leaf photosynthetic parameters using hyperspectral indices in an alpine deciduous forest. Remote Sens, 2020, 12(7): 1124

[32] Kitajima M, Butler WL. Quenching of chlorophyll fluorescence and primary photochemistry in chloroplasts by dibromothymoquinone. Biochim Biophys Acta Bioenerg, 1975, 376(1): 105-115

[33] Kitao M, Lei TT, Koike T, Tobita H, Maruyama Y. Tradeoff between shade adaptation and mitigation of photoinhibition in leaves of Quercus mongolica and Acer mono acclimated to deep shade. Tree Physiol, 2006, 26(4): 441-448

[34] Li JT, Wijewardane NK, Ge YF, Shi YY. Improved chlorophyll and water content estimations at leaf level with a hybrid radiative transfer and machine learning model. Comput Electron Agric, 2023, 206: 107669

[35] Liaw A, Wiener M. Classification and regression by randomForest. R News, 2002, 2: 18-22

[36] Likó SB, Bekő L, Burai P, Holb IJ, Szabó S. Tree species composition mapping with dimension reduction and post-classification using very high-resolution hyperspectral imaging. Sci Rep, 2022, 12: 20919

[37] Liu FS, Hu PC, Zheng BY, Duan T, Zhu BL, Guo Y. A field-based high-throughput method for acquiring canopy architecture using unmanned aerial vehicle images. Agric for Meteorol, 2021, 296: 108231

[38] Liu ZQ, Zhao F, Liu XJ, Yu Q, Wang YF, Peng XB, Cai HJ, Lu XL. Direct estimation of photosynthetic CO₂ assimilation from solar-induced chlorophyll fluorescence (SIF). Remote Sens Environ, 2022, 271: 112893

[39] Lundberg SM, Lee SI (2017) A unified approach to interpreting model predictions. In: Proceedings of the 31st International Conference on Neural Information Processing Systems, Curran Associates Inc., Long Beach, California, USA. pp 4765–4774

[40] Ma Q, Su YJ, Hu TY, Jiang L, Mi XC, Lin LX, Cao M, Wang XG, Lin F, Wang BJ, Sun ZH, Wu J, Ma KP, Guo QH. The coordinated impact of forest internal structural complexity and tree species diversity on forest productivity across forest biomes. Fundam Res, 2024, 4(5): 1185-1195

[41] Mänd P, Hallik L, Peñuelas J, Kull O. Electron transport efficiency at opposite leaf sides: effect of vertical distribution of leaf angle, structure, chlorophyll content and species in a forest canopy. Tree Physiol, 2013, 33(2): 202-210

[42] Mayer M, Watson D (2023) Kernelshap: Kernel SHAP. R package version 0.3 8

[43] Meraner A, Ebel P, Zhu XX, Schmitt M. Cloud removal in Sentinel-2 imagery using a deep residual neural network and SAR-optical data fusion. ISPRS J Photogramm Remote Sens, 2020, 166: 333-346

[44] Mevik BH, Wehrens R, Liland KH (2011) pls: Partial least squares and principal component regression. R package version 2

[45] Meyer D, Dimitriadou E, Hornik K, Weingessel A, Leisch F, Chang CC, Lin CC (2019) e1071: misc functions of the department of statistics, probability theory group (formerly: E1071), TU Wien. R package version 1. https://rdrr.io/rforge/e1071/

[46] Miyake C, Amako K, Shiraishi N, Sugimoto T. Acclimation of tobacco leaves to high light intensity drives the plastoquinone oxidation system: relationship among the fraction of open PSII centers, non-photochemical quenching of Chl fluorescence and the maximum quantum yield of PSII in the dark. Plant Cell Physiol, 2009, 50(4): 730-743

[47] Müller P, Li XP, Niyogi KK. Non-photochemical quenching. A response to excess light energy. Plant Physiol, 2001, 125(4): 1558-1566

[48] Murchie EH, Lawson T. Chlorophyll fluorescence analysis: a guide to good practice and understanding some new applications. J Exp Bot, 2013, 64(13): 3983-3998

[49] Nagai S, Saigusa N, Muraoka H, Nasahara KN. What makes the satellite-based EVI–GPP relationship unclear in a deciduous broad-leaved forest?. Ecol Res, 2010, 25(2): 359-365

[50] Niinemets Ü, Kull O. Sensitivity of photosynthetic electron transport to photoinhibition in a temperate deciduous forest canopy: Photosystem II center openness, non-radiative energy dissipation and excess irradiance under field conditions. Tree Physiol, 2001, 21(12–13): 899-914

[51] Olden JD, Lawler JJ, Poff NL. Machine learning methods without tears: a primer for ecologists. Q Rev Biol, 2008, 83(2): 171-193

[52] Oxborough K, Baker NR. Resolving chlorophyll a fluorescence images of photosynthetic efficiency into photochemical and non-photochemical components–calculation of qP and Fv’/Fm’ without measuring Fo’. Photosynth Res, 1997, 54(2): 135-142

[53] Pollastrini M, Holland V, Brüggemann W, Bruelheide H, Dănilă I, Jaroszewicz B, Valladares F, Bussotti F. Taxonomic and ecological relevance of the chlorophyll a fluorescence signature of tree species in mixed European forests. New Phytol, 2016, 212(1): 51-65

[54] Poorter L, Oberbauer SF, Clark DB. Leaf optical properties along a vertical gradient in a tropical rain forest canopy in Costa Rica. Am J Bot, 1995, 82(10): 1257-1263

[55] Ribeiro RV, Souza GM, Oliveira RF, Machado EC. Photosynthetic responses of tropical tree species from different successional groups under contrasting irradiance conditions. Rev Bras Bot, 2005, 28(1): 149-161

[56] Ryu Y, Berry JA, Baldocchi DD. What is global photosynthesis? History, uncertainties and opportunities. Remote Sens Environ, 2019, 223: 95-114

[57] Scartazza A, Di Baccio D, Bertolotto P, Gavrichkova O, Matteucci G. Investigating the European beech (Fagus sylvatica L.) leaf characteristics along the vertical canopy profile: Leaf structure, photosynthetic capacity, light energy dissipation and photoprotection mechanisms. Tree Physiol, 2016, 36(9): 1060-1076

[58] Shi S, Xu L, Gong W, Chen BW, Chen BW, Qu FF, Tang XT, Sun J, Yang J. A convolution neural network for forest leaf chlorophyll and carotenoid estimation using hyperspectral reflectance. Int J Appl Earth Obs Geoinf, 2022, 108: 102719

[59] Sims DA, Gamon JA. Relationships between leaf pigment content and spectral reflectance across a wide range of species, leaf structures and developmental stages. Remote Sens Environ, 2002, 81(2–3): 337-354

[60] Song GM, Wang Q. Coupling effective variable selection with machine learning techniques for better estimating leaf photosynthetic capacity in a tree species (Fagus crenata Blume) from hyperspectral reflectance. Agric for Meteor, 2023, 338: 109528

[61] Song GM, Wang Q, Zhuang J, Jin J. Timely estimation of leaf chlorophyll fluorescence parameters under varying light regimes by coupling light drivers to leaf traits. Physiol Plant, 2023, 175(5): e14048

[62] Song GM, Wang Q, Zhuang J, Jin J. Dynamics of leaf chlorophyll fluorescence parameters can well be tracked by coupling VIS-NIR-SWIR hyperspectral reflectance and light drivers in partial least-squares regression. Sci Hortic, 2024, 325: 112651

[63] Sonobe R, Wang Q. Assessing hyperspectral indices for tracing chlorophyll fluorescence parameters in deciduous forests. J Environ Manage, 2018, 227: 172-180

[64] Sterling A, Di Rienzo JA. Prediction of South American leaf blight and disease-induced photosynthetic changes in rubber tree, using machine learning techniques on leaf hyperspectral reflectance. Plants, 2022, 11(3): 329

[65] Sun H, Xu ZH, Liu H. An evaluation of the response of vegetation greenness, moisture, fluorescence, and temperature-based remote sensing indicators to drought stress. J Hydrol, 2023, 625: 130125

[66] Turner R, Eriksson D, McCourt M, Kiili J, Laaksonen E, Xu Z, Guyon I (2021) Bayesian optimization is superior to random search for machine learning hyperparameter tuning: analysis of the black-box optimization challenge 2020. arXiv: 2104.10201. https://arxiv.org/abs/2104.10201

[67] Venables WN, Ripley BD. Modern applied statistics with S, 2013, New York. Springer Science & Business Media

[68] Wang Q, Putri NA, Gan Y, Song GM. Combining both spectral and textural indices for alleviating saturation problem in forest LAI estimation using Sentinel-2 data. Geocarto Int, 2022, 37(25): 10511-10531

[69] Wolanin A, Camps-Valls G, Gómez-Chova L, Mateo-García G, van der Tol C, Zhang YG, Guanter L. Estimating crop primary productivity with Sentinel-2 and Landsat 8 using machine learning methods trained with radiative transfer simulations. Remote Sens Environ, 2019, 225: 441-457

[70] Wu Q, Zhang YP, Xie M, Zhao ZW, Yang L, Liu J, Hou DY, Wu Q, Zhang YP, Xie M, Zhao ZW, Yang L, Liu J, Hou DY. Estimation of fv/Fm in spring wheat using UAV-based multispectral and RGB imagery with multiple machine learning methods. Agronomy, 2023, 13(4): 1003

[71] Yang ZF, Tian JC, Wang Z, Feng KP. Monitoring the photosynthetic performance of grape leaves using a hyperspectral-based machine learning model. Eur J Agron, 2022, 140: 126589

[72] Yang QL, Niu CY, Liu XQ, Feng YH, Ma Q, Wang XJ, Tang H, Guo QH. Mapping high-resolution forest aboveground biomass of China using multisource remote sensing data. Gisci Remote Sens, 2023, 60(1): 2203303

[73] Zhang H, Hu H, Zhang XB, Wang KL, Song TQ, Zeng FP. Detecting Suaeda salsa L. chlorophyll fluorescence response to salinity stress by using hyperspectral reflectance. Acta Physiol Plant, 2012, 34(2): 581-588

[74] Zhang JR, Xiao JF, Tong XJ, Zhang JS, Meng P, Li J, Liu PR, Yu PY. NIRv and SIF better estimate phenology than NDVI and EVI: effects of spring and autumn phenology on ecosystem production of planted forests. Agric for Meteor, 2022, 315: 108819

[75] Zhao RM, An LL, Song D, Li MZ, Qiao L, Liu N, Sun H. Detection of chlorophyll fluorescence parameters of potato leaves based on continuous wavelet transform and spectral analysis. Spectrochim Acta A Mol Biomol Spectrosc, 2021, 259: 119768

[76] Zhao X, Qi JB, Yu ZX, Yuan LJ, Huang HG. Fine-scale quantification of absorbed photosynthetically active radiation (APAR) in plantation forests with 3D radiative transfer modeling and LiDAR data. Plant Phenomics, 2024, 6: 166

[77] Zheng C, Wang SQ, Chen JM, Chen JH, Chen B, He XL, Li H, Sun LG. Combination of vegetation indices and SIF can better track phenological metrics and gross primary production. JGR Biogeosci, 2023, 128(7): e2022JG007315

[78] Zhuang J, Wang Q. Within-canopy variation in chlorophyll fluorescence parameters can be well captured by vertical patterns of leaf biophysical and chemical traits. Physiol Plant, 2024, 176(4): e14462

[79] Zhuang J, Wang Q, Song GM, Jin J, Zhuang J, Wang Q, Song GM, Jin J. Validating and developing hyperspectral indices for tracing leaf chlorophyll fluorescence parameters under varying light conditions. Remote Sens, 2023, 15(19): 4890

RIGHTS & PERMISSIONS

The Author(s)