Search

In order to explore the potential of L-band full-polarization SAR data to estimate forest aboveground biomass (AGB), five polarimetric scattering ratio parameters (R1, R2, R3, R4, R5) were constructed based on the canopy-ground scattering component of Unmmaned Aero Vehicle Synthetic Aperture Radar (UAVSAR) data of the AfricaSAR project. Calculating the Radar Vegetation Index (RVI), and 21 polarimetric decomposition parameters were extracted by four model-based decompositions, including six-component and seven-component decomposition. Finally, all features were merged and the random forest feature importance was used to screen out the optimal feature combination, and random forest (RF), support vector machine regression (SVR), K-nearest neighbor regression (KNN) were used to estimate forest AGB of Lope, The Gaboneses Repbulic, Africa, with different feature combinations. The results showed that the polarimetric scattering ratio parameters, bulk scattering (Vol) and RVI had high sensitivity to forest AGB, and the correlation between R2 and AGB was 0.823, and the optimal feature combination was Vol, polarimetric scattering ratio parameters and RVI. Machine learning models with different feature combinations had shown good performance, the coefficient of determination (R²) of the machine learning model based on the polarimetric decomposition parameters was bigger than 0.800, and the root mean square error (RMSE) was less than 88.000 Mg/hm², and the best effect was the RF model based on the optimal feature combination, which increased R² by 0.144 and decreased RMSE by 30.327 Mg/hm² compared with the polarimetric decomposition parameters alone. The polarimetric scattering ratio parameter had certain potential in the estimation of forest AGB, the introduction of RVI improved the accuracy of the model, the model-based decomposition was suitable for forest AGB estimation, and the machine learning model based on feature screening can better invert forest AGB, and there was no obvious saturation point when the AGB reached 400 Mg/hm². [ABSTRACT FROM AUTHOR]

In order to explore the advantages and complementarity of optical data and synthetic aperture radar( SAR) data in for‐ est type classification, this study focused on the overlapping area of Landsat8 data and ALOS2 data from one scene SAR image in Simao District, Puer City, Yunnan Province, China, and used hierarchical classification technology for forest type classification re‐ search. Three feature sets were constructed： optical feature set( spectral + vegetation + texture + terrain features), SAR feature set( backscattering + polarization decomposition features), and optical-SAR fusion feature set( spectral + vegetation + texture + terrain + backscattering + polarization decomposition features) . Recursive Feature Elimination( RFE) was employed to perform stratified fea‐ ture selection on the three feature sets, and random forest( RF) and support vector machine( SVM) were used for forest type classifi‐ cation. The SVM classification with the fusion of optical images and SAR data achieved the best results. The results showed,1) In the first layer( vegetation/non-vegetation) classification, the overall accuracy was 98. 57%, the Kappa coefficient was 0. 971. 2) In the second layer( forest/non-forest) classification, the overall accuracy was 92. 14%, the Kappa coefficient was 0. 826. 3) In the third layer( coniferous/broad-leaved/mixed forest) classification, the overall accuracy was 83. 47%, and the Kappa coefficient was 0. 743. The fusion data showed an improvement of 6. 74% in accuracy compared to optical data feature set classification and 29. 24% com‐ pared to SAR classification. 4) In the classification of the third layer using fusion data, the influence of different window sizes( 3×3, 5×5,7×7,9×9) of texture features in optical images was compared, and the highest accuracy was achieved with a 7×7 texture win‐ dow. Results shows that, the accuracy of forest type classification using multi-source data is higher than that using a single data source. [ABSTRACT FROM AUTHOR]

Quickly and accurately obtaining land use information can provide a reference for urban development and ecological environment protection. This study classifies land use types in Yunnan Province based on multi-temporal Landsat image dense time stacking and random forest algorithm on the Google Earth Engine （GEE） platform, analyzes the spatiotemporal change trends of LULC in Yunnan Province, and uses geodetectors to quantitatively evaluate key drivers factor. The results show that, 1） the average overall accuracy and Kappa coefficient of LULC classification in this study are 88. 64% and 86. 01%, respectively, which is highly accurate and meets the data usage requirements. 2） The land types in Yunnan Province are mainly forest land, cultivated land, grassland, and sparse shrub grass mixed land, accounting for 97. 91%-98. 38%. Land use transfer mainly involves the conversion of forest land and cultivated land, and the conversion of grassland and sparse shrub grass mixed land into cultivated land. 3） The land use intensity in central and eastern Yunnan in Yunnan Province is generally higher than that in other regions, while the land use intensity in northwest and southwestern Yunnan is lower. 4） There are significant differences in the influence of different driving factors on LULC. Vegeta‐ tion type, average annual temperature and soil type have a relatively small impact on LULC changes. Elevation, slope, aspect, aver‐ age annual precipitation, population density, GDP and population urbanization rate, etc., generally have a high impact on LULC changes, Among them, GDP, population density, and population urbanization rate have a higher impact on LULC changes. The re‐ search results can provide data basis and support for subsequent ecological and environmental protection policy formulation and re‐ gional sustainable development in Yunnan Province. [ABSTRACT FROM AUTHOR]

Copyright of Forest Research is the property of Forest Research Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Forest Research is the property of Forest Research Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

【Objective】The present papar aims to discus on integrated optical and microwave remote sensing inversion methods of forest volume from the data source. 【Method】Based on ALOS PALSAR L-band quad-polarimetric microwave remote sensing data and Landsat TM multi-spectral optical remote sensing data, quad-polarimetric backscattering coefficients, polarization ratio and optical factors were extracted with the coordinates of forest sample plots, and forest volume inversion models were built with multiple linear regression. 【Result】The results showed that three inversion methods, conventional optical data, microwave data and the multi-source data combined former two data source could all reach the objective to estimate forest volume, especially the multi-source inversion model, it had better results with R2 of 0.674 and RMSEof 13.38 m3/hm2. 【Conclusion】Compared with the inversion model using single data source, multi-source inversion model had obvious advantages, and effectively realized forest volume estimation. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Remote Sensing is the property of Editorial Office of Journal of Remote Sensing & Science Publishing Co. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)