Development of Four Component Scattering Power Decomposition Technique for Dual Polarization SAR Data

Journal of the Indian Society of Remote Sensing, Год журнала: 2024, Номер unknown

Опубликована: Ноя. 14, 2024

Abstract The increasing availability of dual-polarimetric synthetic aperture radar (PolSAR) data has led to a significant rise in its applications over the past few decades. Model-based decompositions combined with polarimetric information extraction from PolSAR play crucial role target identification and classification. In this context, covariance matrix [C], composed four independent parameters, was used as input for dual-pol four-component scattering power decomposition (DP-4SD). A novel 4SD model tested using dual SAR spaceborne ALOS-2/PALSAR-2, performance evaluated against existing methods. proposed assessed dual-polarization Haldwani Forest San Francisco evaluate classification capabilities within single class (forest) across various land use cover classes Francisco. overall accuracy achieved 85.69% forest 93.66% Francisco, fewer unclassified samples compared model. demonstrates superior enhances interpretation information, indicating potential significantly improve land-use land-cover mapping data.

Язык: Английский

Studying Forest Species Classification Methods by Combining PolSAR and Vegetation Spectral Indices

Forests, Год журнала: 2024, Номер 16(1), С. 15 - 15

Опубликована: Дек. 25, 2024

Tree species are important factors affecting the carbon sequestration capacity of forests and maintaining stability ecosystems, but trees widely distributed spatially located in complex environments, there is a lack large-scale regional tree classification models for remote sensing imagery. Therefore, many studies aim to solve this problem by combining multivariate data proposing machine learning model forest classification. However, satellite-based laser systems find it difficult meet needs characters, due their unique footprint sampling method, SAR limit accuracy classification, information blending backscatter coefficients. In work, we combined Sentinel-1 Sentinel-2 construct based on optical features, vegetation spectral PolSAR polarization observation propose feature selection method featuring Hilbert–Huang transform mixed surface data. The PSO-RF was used classify species, including four temperate broadleaf forests, namely, aspen (Populus L.), maple (Acer), peach (Prunus persica), apricot armeniaca two coniferous Chinese pine (Pinus tabuliformis Carrière) Mongolian sylvestris var. mongolica Litv.). study, some experiments were conducted using images, 550 measured survey sample points pertaining forested area Fuxin District, Liaoning Province, China. results show that fusion constructed study has high accuracy, with Kappa coefficient 0.94 an overall 95.1%. addition, shows can play role applying data, other interferes perceived vertical structure be suppressed certain extent, its PolSAR, should not ignored.

Язык: Английский