当前位置: X-MOL 学术Remote Sens. Environ. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Identification of saline landscapes from an integrated SVM approach from a novel 3-D classification schema using Sentinel-1 dual-polarized SAR data
Remote Sensing of Environment ( IF 11.1 ) Pub Date : 2022-06-28 , DOI: 10.1016/j.rse.2022.113144
Shoba Periasamy , Kokila Priya Ravi , Kevin Tansey

This study presented an integrated SVM classification method that investigated the relationship between intrinsic scattering attributes of the surface features and the conductivity characteristics of the soil in the newly proposed two- and three-dimensional classification (2D and 3D) schema to classify the saline landscape. The study was conducted in the Vellore district, Tamil Nadu, which is composed of heterogeneous saline landforms due to the active geological processes and anthropogenic activities. The intensity products of Sentinel-1 data (VV + VH) of C-band frequency were employed in the present study. The SVM with linear kernel has outperformed the other models, namely random forest (RF), naive Bayes (NB), and K-Nearest Neighbour (k−NN), in performing the broad level of classification from the 2D classification schema (SVMOA = 84.2%, RFOA = 80.4%, k-NNOA = 78.8%, NBOA = 68.4%). The soil EC values derived from the dielectric loss measurements (R2 = 0.79, ρ = 0.018, and α=95%) were used to introduce the integrated SVM approach in the 3D schema to further breakdown the mapped classes into non-saline (NS) (soil EC ≤ 2 ds/cm), slightly saline (SS) (soil EC = 2.1 to 4 ds/cm) and moderately saline (MS) (soil EC = 4.1 to 8 ds/cm) categories. The overall performance of the integrated SVM approach implemented for the 3D classification schema (F1 = 0.80) was found to be satisfactory, but with an associated uncertainty majorly from MS (Precision = 0.52, F1 = 0.69), SS (Recall = 0.09, F1 = 0.15), and NS waterbodies (Recall = 0.18, F1 = 0.29) as shown in the Precision-Recall graph (AUCPR3D = 0.62). However, with the promising performance level demonstrated for the other nine classes such as NS, SS, and MS wet soil (F1 = 0.92, 0.92, 0.96), healthy plants (F1 = 0.83), salt-tolerant plants under SS and MS conditions (F1 = 0.83, 0.88), and waterlogged vegetation under NS, SS, and MS conditions (F1 = 0.82, 0.83, 0.83), the proposed classification scheme becomes an effective method to map saline and non-saline features from dual-polarimetric SAR data.



中文翻译:

使用 Sentinel-1 双极化 SAR 数据从新的 3-D 分类模式中通过集成 SVM 方法识别盐碱地景观

本研究提出了一种集成的 SVM 分类方法,该方法研究了新提出的二维和三维分类(2D 和 3D)模式中地表特征的固有散射属性与土壤的电导率特征之间的关系,以对盐碱地进行分类。该研究是在泰米尔纳德邦的韦洛尔地区进行的,由于活跃的地质过程和人为活动,该地区由异质盐碱地貌组成。本研究采用 C 波段频率的 Sentinel-1 数据 (VV + VH) 的强度乘积。具有线性核的 SVM 的性能优于其他模型,即随机森林 (RF)、朴素贝叶斯 (NB) 和 K-Nearest Neighbor ( k−NN),在从 2D 分类模式执行广泛级别的分类时(SVM OA  = 84.2%,RF OA  = 80.4%,k-NN OA  = 78.8%,NB OA  = 68.4%)。从介电损耗测量得出的土壤 EC 值(R 2  = 0.79、ρ  = 0.018 和α=95%) 用于在 3D 模式中引入集成 SVM 方法,以进一步将映射的类别分解为非盐 (NS) (土壤 EC ≤ 2 ds/cm)、微盐 (SS) (土壤 EC = 2.1 到4 ds/cm) 和中度盐碱 (MS) (土壤 EC = 4.1 至 8 ds/cm) 类别。发现为 3D 分类模式 (F1 = 0.80) 实施的集成 SVM 方法的整体性能令人满意,但相关的不确定性主要来自 MS (Precision = 0.52, F1 = 0.69), SS (Recall = 0.09, F1 = 0.15) 和 NS 水体 (Recall = 0.18, F1 = 0.29),如 Precision-Recall 图 (AUCPR 3D = 0.62)。然而,其他九类如 NS、SS 和 MS 湿土 (F1 = 0.92, 0.92, 0.96)、健康植物 (F1 = 0.83)、SS 和 MS 条件下的耐盐植物等表现出有希望的性能水平(F1 = 0.83, 0.88) 和 NS、SS 和 MS 条件下的淹水植被 (F1 = 0.82, 0.83, 0.83),所提出的分类方案成为从双极化 SAR 中绘制咸水和非咸水特征的有效方法数据。

更新日期:2022-06-28
down
wechat
bug