In order to accurately obtain forest canopy height information, accurately estimate forest aboveground biomass, and evaluate forest carbon sink capacity, this study constructed 30 long time series feature variables based on ground measurements, multi-source remote sensing data, and digital elevation models, combined with machine learning algorithms (ML), to invert the forest canopy height in Lishui City, Zhejiang Province. The study revealed that terrain factors had no significant impact on the inversion of forest canopy height, while vegetation factors related to the red and green bands were strongly correlated with forest canopy height. Adding long time series feature factors can help improve the accuracy of ML algorithm in inverting forest canopy height. The performance improvement of CNN was particularly significant, achieving an optimal coefficient of determination (R2) increase of 0.39 and a root mean square error (RMSE in the formula, it is denoted as RMES) decrease of 4.15. Random forest had the highest inversion accuracy among the four ML algorithms (R2=0.79, RMSE=1.65), greater than support vector machine (R2=0.65, RMSE=1.97), extreme gradient ascent method (R2=0.76, RMSE=1.81) and convolutional neural networks (R2=0.71, RMSE=1.83).
LEFSKYM A, HARDINGD J, KELLERM,et al.Estimates of forest canopy height and aboveground biomass using ICESat[J].Geophysical Research Letters,2005,32(22):-L22S02.
[2]
TAOS, GUOQ, LIC,et al.Global patterns and determinants of forest canopy height[J].Ecology:A Publication of the Ecological Society of America,2016,97(12):3265-3270.
[3]
KLEINT, RANDINC, KÖRNERC.Water availability predicts forest canopy height at the global scale[J].Ecology Letters,2015,18(12):1311-1320.
[4]
PASCUALC, GARCÍA-ABRILA, COHENW B,et al.Relationship between LiDAR-derived forest canopy height and Landsat images[J].International Journal of Remote Sensing,2010,31(5):1261-1280.
[5]
MAOL, BATERC W, STADTJ J,et al.Environmental landscape determinants of maximum forest canopy height of boreal forests[J].Journal of Plant Ecology,2019,12(1) 96-102.
[6]
SIMARDM, PINTON, FISHERJ B,et al.Mapping forest canopy height globally with spaceborne lidar[J].Journal of Geophysical Research,2011,116(G4):G04021.
[7]
ZHAOJ, ZHAOL, CHENE,et al.An improved generalized hierarchical estimation framework with geostatistics for mapping forest parameters and its uncertainty:a case study of forest canopy height[J].Remote Sensing,2022,14(3):568.
XIEK Y, CHENR B, WANGZ L,et al.The inversion and application of forest height of multi-source remote sensing data in Guangxi-ASEAN region[J].Bulletin of Surveying and Mapping,2024(1):32-37,64.
LIUS T, WANGX M, ZHAOF.Estimation of aboveground vegetation biomass in oasis-desert transition zone based on Sentinel-2[J].Journal of Arid Land Resources and Environment,2024,38(4):162-170
DUANY F, LUOH B, YUEC R.Application UAVSAR data and improved polarized water cloud model for aboveground biomass in tropical forests[J].Journal of Northeast Forestry University,2024,52(1):54-60.
LIUS K, SHIZ L, SONGH Y,et al.Effects of slope on the estimates of individual tree parameters for coniferous plantation using airborne laser scanner[J].Journal of Northeast Forestry University,2021,49(4):45-51.
[18]
TRAVERS-SMITHH, COOPSN C, MULVERHILLC,et al.Mapping vegetation height and identifying the northern forest limit across Canada using ICESat-2,Landsat time series and topographic data[J].Remote Sensing of Environment,2024,305:114097.
[19]
VERASH F P, FERREIRAM P, NETOE M D C,et al.Fusing multi-season UAS images with convolutional neural networks to map tree species in Amazonian forests[J].Ecological Informatics,2022,71:101815.
[20]
TUNN L, GAVRILOVA, TUNN M,et al.Remote sensing data classification using a hybrid pre-trained VGG16 CNN-SVM classifier[C]//2021 IEEE Conference of Russian Young Researchers in Electrical and Electronic Engineering:ElConRus,2021:2171-2175.
CAIL F, WUD S, FANGL M,et al.Tree species identification using XGBoost based on GF-2 images[J] Forestry Resource Management,2019(5):44-51.
[23]
FAYADI, CIAISP, SCHWARTZM,et al.Hy-TeC:a hybrid vision transformer model for high-resolution and large-scale mapping of canopy height[J].Remote Sensing of Environment,2024,302:113945.
[24]
WANGZ, ZOUC, CAIW.Small sample classification of hyperspectral remote sensing images based on sequential joint deeping learning model[J].IEEE Access,2020,8:71353-71363.
[25]
GUOX, LIH, JINGL,et al.Individual tree species classification based on convolutional neural networks and multitemporal high-resolution remote sensing images[J].Sensors (Basel,Switzerland),2022,22(9):3157.
[26]
TAMIMINIAH, SALEHIB, MAHDIANPARIM,et al.State-wide forest canopy height and aboveground biomass map for New York with 10 m resolution,integrating GEDI,Sentinel-1,and Sentinel-2 data[J].Ecological Informatics,2024,79:102404.
[27]
ZHUW, LIY, LUANK,et al.Forest canopy height retrieval and analysis using random forest model with multi-source remote sensing integration[J].Sustainability,2024,16(5):1-21.
[28]
GUANY, TIANX, ZHANGW,et al.Forest canopy cover inversion exploration using multi-source optical data and combined methods[J].Forests,2023,14(8):1527.
[29]
SCHLERFM, ATZBERGERC.Inversion of a forest reflectance model to estimate structural canopy variables from hyperspectral remote sensing data[J].Remote Sensing of Environment,2006,100(3):281-294.
HEJ Y, JIAWW, ZHANGX Y,et al.Remote sensing estimation of forest canopy LAI using different algorithms of PROSAIL model[J].Journal of Northeast Forestry University,2023,51(11):86-94.
[32]
MAH, SONGJ, WANGJ,et al.Comparison of the inversion ability in extrapolating forest canopy height by integration of LiDAR data and different optical remote sensing products[C]//2012 IEEE International Geoscience and Remote Sensing Symposium.2012:3363-3366.
[33]
ZHANGQ, GEL, HENSLEYS,et al.PolGAN:a deep-learning-based unsupervised forest height estimation based on the synergy of PolInSAR and LiDAR data[J].ISPRS Journal of Photogrammetry and Remote Sensing,2022,186:123-139.