Leaf area index estimation in the Zagros forests of Iran using Sentinel-2 image and Gaussian Process Regression

Document Type : Research article

Authors

1 Ph.D. Student, Department of Forestry and Forest Economics, Faculty of Natural Resources, University of Tehran, Karaj, Iran

2 Assistant Prof., Department of Forestry and Forest Economics, Faculty of Natural Resources, University of Tehran, Karaj, Iran

3 Prof., Department of Forestry and Forest Economics, Faculty of Natural Resources, University of Tehran, Karaj, Iran

4 Associate Prof., Department of Forestry, Academic Member of Dr. Hedayat Ghazanfari Center for Research & Development of Northern Zagros Forestry, University of Kurdistan, Sanandaj, Iran

5 Assistant Prof., Department of Remote Sensing, Global Change Research Institute of the Czech Academy of Sciences, Brno, Czech Republic

10.22092/ijfpr.2023.364041.2129

Abstract

Background and objectives: Leaf area index (LAI) is a vital biophysical characteristic to assess the condition, describe forest structure and function of forest ecosystems. LAI is a key input in modeling global climate change, carbon fluxes, water cycle, photosynthesis, and interception processes. The estimation of LAI in forests through remote sensing data, using machine learning models, has gained widespread attention, particularly for large-scale LAI mapping. This method is favored for its efficiency, involving minimal time investment, cost-effectiveness, and a non-destructive approach. This study aimed to investigate the potential of Sentinel-2 data for estimating the LAI of northern Zagros forests, employing the Gaussian Process Regression (GPR) method.
Methodology: LAI field data were collected in June and July 2023 from a coppice forest in the Marivan and Sarvabad counties of Kurdistan province, Iran. A total of 93 square plots, each measuring 20×20 square meters, were randomly selected. The location of each plot was recorded using a DGPS device. The LAI within each plot was measured using the hemispherical photography method. Five photos were captured within each sample using a Coolpix4500+FC-E8 camera equipped with a fisheye lens. The LAI was then calculated for each hemispherical photo and averaged for each sample plot using the “hemispheR” package in the R programming language. A cloud-free Sentinel-2B image with L1C correction level was acquired on July 2, 2023. After verifying the radiometric and geometric quality of the image, the Sen2Cor processor was used to apply atmospheric correction. Different input data, including spectral bands and spectral indices (Vegetation Indices, Tasseled Cap Transformation, and Principal Component Analysis) were generated from the Sentinel-2 image. These datasets, i.e., the spectral bands, spectral indices, and a combination of spectral bands and spectral indices, were used to estimate LAI. The modeling process was carried out using the GPR algorithm based on 65 sample plots (70% of the dataset). The performance of the models was finally evaluated using 28 plots (30% of the dataset) with different metrics such as the coefficient of determination (R2), root mean square error (RMSE), relative root mean square error (rRMSE), and Akaike Information Criterion (AIC).
Results: The descriptive statistics for the measured LAI showed that the minimum, maximum, average, and standard deviation values of the leaf area index over the study area were 0.33, 3.88, 2.129, and 0.627 m2.m-2, respectively. The Pearson correlation analysis between forest LAI and spectral variables (including original bands and spectral indices) indicated a stronger correlation between LAI and spectral indices (i.e., GNDVI, SAVI, and TCTV) than the original bands. Thirty percent of field sample plots were randomly selected and used to evaluate the forest LAI model generated using the GPR machine learning algorithm based on three datasets: original bands, spectral indices, and a combination of original bands and spectral indices, all derived from Sentinel-2 imagery. The evaluation outcomes revealed that the model derived from the main bands of the Sentinel-2 satellite achieved R2 = 0.81, RMSE = 0.21 m2.m-2, rRMSE = 9.14%, and AIC = 103.65. This performance was deemed satisfactory when compared to the performance of models built using the other two datasets (i.e., spectral indices, and a combination of original bands and spectral indices) to estimate LAI. Using the best-performing model, a comprehensive LAI map of the study area was generated using data derived from the main bands of Sentinel-2 imagery.
Conclusion: This study provides preliminary evidence of the potential of Sentinel-2 satellite data in evaluating the leaf area index in the North Zagros coppice forests. However, the integration of ground data of leaf area index and Sentinel-2 data from various growing seasons could potentially enhance the robustness of the results and mitigate uncertainties, thereby paving the way for future research endeavors. This approach could lead to more accurate and reliable assessments of forest health and productivity.
 
 

Keywords

Main Subjects


- Ali, A.M., Darvishzadeh, R., Skidmore, A., Gara, T.W. and Heurich, M., 2021. Machine learning methods’ performance in radiative transfer model inversion to retrieve plant traits from Sentinel-2 data of a mixed mountain forest. International Journal of Digital Earth, 14(1): 106-120.‏
- Bhattarai, R., Rahimzadeh-Bajgiran, P., Weiskittel, A., Homayouni, S., Gara, T.W. and Hanavan, R.P., 2022. Estimating species-specific leaf area index and basal area using optical and SAR remote sensing data in Acadian mixed spruce-fir forests, USA. International Journal of Applied Earth Observation and Geoinformation, 108: 102727.‏
- Campos-Taberner, M., García-Haro, F.J., Busetto, L., Ranghetti, L., Martínez, B., Gilabert, M.A., ... and Boschetti, M., 2018. A critical comparison of remote sensing Leaf Area Index estimates over rice-cultivated areas: From Sentinel-2 and Landsat-7/8 to MODIS, GEOV1 and EUMETSAT polar system. Remote Sensing, 10(5): 763.‏
- Chen, Z., Jia, K., Xiao, C., Wei, D., Zhao, X., Lan, J., ... and Wang, L., 2020. Leaf area index estimation algorithm for GF-5 hyperspectral data based on different feature selection and machine learning methods. Remote Sensing, 12(13): 2110.‏
- Chianucci, F. and Macek, M., 2023. hemispheR: an R package for fisheye canopy image analysis. Agricultural and Forest Meteorology, 336: 109470.‏
- Chrysafis, I., Korakis, G., Kyriazopoulos, A.P. and Mallinis, G., 2020. Retrieval of leaf area index using Sentinel-2 imagery in a mixed Mediterranean forest area. ISPRS International Journal of Geo-Information, 9(11): 622.‏
- Cui, S. and Zhou, K., 2017. A comparison of the predictive potential of various vegetation indices for leaf chlorophyll content. Earth Science Informatics, 10(2): 169-181.‏
- Darvishzadeh, R., Skidmore, A., Abdullah, H., Cherenet, E., Ali, A., Wang, T., ... and Paganini, M., 2019. Mapping leaf chlorophyll content from Sentinel-2 and RapidEye data in spruce stands using the invertible forest reflectance model. International Journal of Applied Earth Observation and Geoinformation, 79: 58-70.‏
- Dube, T., Pandit, S., Shoko, C., Ramoelo, A., Mazvimavi, D. and Dalu, T., 2019. Numerical assessments of leaf area index in tropical savanna rangelands, South Africa using Landsat 8 OLI derived metrics and in-situ measurements. Remote Sensing, 11(7): 829.‏
- Erfanifard, Y. and Lotfi Nasirabad, M., 2022. Comparison of vegetation and mangrove indices in mangrove mapping on Sentinel-2 imagery based on Google Earth Engine. Iranian Journal of Forest and Poplar Research, 30(3): 224-240 (In Persian with English summary).
- Estévez, J., Salinero-Delgado, M., Berger, K., Pipia, L., Rivera-Caicedo, J.P., Wocher, M., ... and Verrelst, J., 2022. Gaussian processes retrieval of crop traits in Google Earth Engine based on Sentinel-2 top-of-atmosphere data. Remote Sensing of Environment, 273: 112958.‏
- Fallah, A., Nazariani, N., Imani Rastabi, M., Bakhshi, F., 2022. Modeling the commercial volume of pure and mixed stands of beech trees using non-parametric algorithms in the educational-research Forest of Darabkola, Sari, Iran. Iranian Journal of Forest and Poplar Research, 30(2): 180-192 (In Persian with English summary).
- Gewali, U.B., Monteiro, S.T. and Saber, E., 2019. Gaussian processes for vegetation parameter estimation from hyperspectral data with limited ground truth. Remote Sensing, 11(13): 1614.‏
- Jonckheere, I., Fleck, S., Nackaerts, K., Muys, B., Coppin, P., Weiss, M. and Baret, F., 2004. Review of methods for in situ leaf area index determination: Part I. Theories, sensors and hemispherical photography. Agricultural and Forest Meteorology, 121(1-2): 19-35.‏
- Kovacs, J.M., Flores-Verdugo, F., Wang, J. and Aspden, L.P., 2004. Estimating leaf area index of a degraded mangrove forest using high spatial resolution satellite data. Aquatic Botany, 80(1): 13-22.
- Mao, H., Meng, J., Ji, F., Zhang, Q. and Fang, H., 2019. Comparison of machine learning regression algorithms for cotton leaf area index retrieval using Sentinel-2 spectral bands. Applied Sciences, 9(7): 1459.‏
- Meyer, L.H., Heurich, M., Beudert, B., Premier, J. and Pflugmacher, D., 2019. Comparison of Landsat-8 and Sentinel-2 data for estimation of leaf area index in temperate forests. Remote Sensing, 11(10): 1160.‏
- Miri, N., Darvishsefet, A.A., Zargham, N. and Shakeri, Z., 2017. Estimation of leaf area index in Zagros forests using Landsat 8 data. Iranian Journal of Forest, 9(1): 29-42 (In Persian with English summary).
- Moradi, G., Pir Bavaghar, M., Shakeri, Z. and Fatehi, P., 2021. Leaf area index estimation in the northern Zagros forests using remote sensing (Case study: a part of Baneh forests). Journal of Forest Research and Development, 6(4): 679-693 (In Persian with English summary).
- Omer, G., Mutanga, O., Abdel-Rahman, E.M. and Adam, E., 2016. Empirical prediction of leaf area index (LAI) of endangered tree species in intact and fragmented indigenous forests ecosystems using WorldView-2 data and two robust machine learning algorithms. Remote Sensing, 8(4): 324.‏
- Pope, G. and Treitz, P., 2013. Leaf area index (LAI) estimation in boreal mixedwood forest of Ontario, Canada using light detection and ranging (LiDAR) and WorldView-2 imagery. Remote Sensing, 5(10): 5040-5063.‏
- Sinha, S.K., Padalia, H., Dasgupta, A., Verrelst, J. and Rivera, J.P., 2020. Estimation of leaf area index using PROSAIL based LUT inversion, MLRA-GPR and empirical models: Case study of tropical deciduous forest plantation, North India. International Journal of Applied Earth Observation and Geoinformation, 86: 102027.‏
- Verrelst, J., Alonso, L., Caicedo, J.P.R., Moreno, J. and Camps-Valls, G., 2013. Gaussian process retrieval of chlorophyll content from imaging spectroscopy data. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 6(2): 867-874.‏
- Verrelst, J., Malenovský, Z., Van der Tol, C., Camps-Valls, G., Gastellu-Etchegorry, J.P., Lewis, P., ... and Moreno, J., 2019. Quantifying vegetation biophysical variables from imaging spectroscopy data: a review on retrieval methods. Surveys in Geophysics, 40: 589-629.‏
- Verrelst, J., Rivera, J.P., Veroustraete, F., Muñoz-Marí, J., Clevers, J.G.P.W., Camps-Valls, G. and Moreno, J., 2015. Experimental Sentinel-2 LAI estimation using parametric, non-parametric and physical retrieval methods – A comparison. ISPRS Journal of Photogrammetry and Remote Sensing, 108: 260-272.‏
- Weiss, M., Baret, F., Smith, G.J., Jonckheere, I. and Coppin, P., 2004. Review of methods for in situ leaf area index (LAI) determination: Part II. Estimation of LAI, errors and sampling. Agricultural and Forest Meteorology, 121(1-2): 37-53.‏
- Wocher, M., Berger, K., Verrelst, J. and Hank, T., 2022. Retrieval of carbon content and biomass from hyperspectral imagery over cultivated areas. ISPRS Journal of Photogrammetry and Remote Sensing, 193: 104-114.‏
- Xie, R., Darvishzadeh, R., Skidmore, A.K., Heurich, M., Holzwarth, S., Gara, T.W. and Reusen, I., 2021. Mapping leaf area index in a mixed temperate forest using Fenix airborne hyperspectral data and Gaussian processes regression. International Journal of Applied Earth Observation and Geoinformation, 95: 102242.‏
- Xu, J., Quackenbush, L.J., Volk, T.A. and Im, J., 2020. Forest and crop leaf area index estimation using remote sensing: Research trends and future directions. Remote Sensing, 12(18): 2934.‏
- Zhang, F., Tian, X., Zhang, H. and Jiang, M., 2022. Estimation of aboveground carbon density of forests using deep learning and multisource remote sensing. Remote Sensing, 14(13): 3022.‏
- Zou, J., Hou, W., Chen, L., Wang, Q., Zhong, P., Zuo, Y., ... and Leng, P., 2020. Evaluating the impact of sampling schemes on leaf area index measurements from digital hemispherical photography in Larix principis-rupprechtii forest plots. Forest Ecosystems, 7(1): 52.‏
- Zou, X., Zhu, S. and Mõttus, M., 2022. Estimation of canopy structure of field crops using sentinel-2 bands with vegetation indices and machine learning algorithms. Remote Sensing, 14(12): 2849.‏