A Comparison of Gaofen-2 and Sentinel-2 Imagery for Mapping Mangrove Forests Using Object-Oriented Analysis and Random Forest

Abstract

Mangrove forest extents and distributions are fundamental for conservation and restoration efforts. According to previous studies, both the commercial Gaofen-2 (GF-2) imagery (0.8 m spatial resolution and 4 spectral bands) and freely accessed Sentinel-2 (S2) imagery (10 m spatial resolution and 13 spectral bands) have been successfully used to map mangrove forests. However, the efficiency and accuracy of mangrove forest mapping based on these two data is not clear, especially for large-scale applications. To address this issue, firstly, we developed a robust classification approach by integrating object-based image analysis (OBIA) and random forest (RF) algorithm; and then, applied this approach to GF-2 and S2 images to map the extents of mangrove forest along the entire coasts of Guangxi, China, respectively; at last, compared the efficiency and accuracy of GF-2 and S2 imagery in mangrove forest mapping. Results showed that: (1) based on OBIA and RF integrated classification approach both mangrove forest maps derived from GF-2 and S2 obtained high mapping accuracies (the overall accuracy was 96% and 94%, respectively); (2) areal extent of mangrove forests in Guangxi extracted from GF-2 and S2 images was 8182 ha and 8040 ha, respectively; (3) GF-2 imagery has extraordinary abilities in detecting fragmented mangrove forest patches located along landward and seaward edges; (4) S2 imagery performed better in detecting seaward submerged mangrove forests and separating mangrove forest from terrestrial vegetation. Results and conclusions of this study can provide basic considerations for selecting appropriate data source in mangrove forest or wetland vegetation mapping tasks.

Quantifying the Long-Term Expansion and Dieback of Spartina Alterniflora Using Google Earth Engine and Object-Based Hierarchical Random Forest Classification

“…Image segmentation parameters include scale, shape, color, smoothness, and compactness. These five parameters determine the scale of the segmentation effect [27]. The scale determines heterogeneity of the image objects, with large scale parameters producing large image objects and vice versa.…”

“…As an improved technique, OBIA aims to segment the pixels (with the same attributes) into objects for classification according to certain rules [25], [26]. OBIA has been successfully used in many Spartina saltmarsh mappings [9], [27]- [29]. For example, Liu (2018) used OBIA and SVM methods to quantify the dynamics of S. alterniflora in China with high overall accuracy [9].…”

“…For example, Liu (2018) used OBIA and SVM methods to quantify the dynamics of S. alterniflora in China with high overall accuracy [9]. However, Some studies have focused on combining OBIA with an classifier, i.e., focusing on a year or multiple years to study the dynamics of S. alterniflora based on image quality [9], [27], [30], [31]. Given their medium spatial resolution, wide coverage, and consistent availability, Landsat images have been most widely used to describe the characteristics of invasive plants since the 1980s [1], [9], [11].…”

Mangrove ecosystem species mapping from integrated Sentinel-2 imagery and field spectral data using random forest algorithm

“…The ultimate prediction in the RF method was created by voting together the predictions of many decision trees. 4 For classification tasks, the majority vote was considered the final prediction. In classification, to improve processing effectiveness, it is crucial to understand how each variable affects the outcomes.…”

Digital Soil Mapping Based on Fine Temporal Resolution Landsat Data Produced by Spatiotemporal Fusion

“…Owing to the random and bootstrap-based sampling technique, the RF model can effectively hinder overfitting. Therefore, many studies have adopted the RF model for regression and classification [47]- [49]. As a key parameter, the number of trees (n_tree) should be optimized by testing the out-of-bag (OOB) error in the RF model.…”