Mapping and Change Analysis in Mangrove Forest by Using Landsat Imagery

Abstract

ABSTRACT:Mangrove is located in the tropical and subtropical regions and brings good services for native people. Mangrove in the world has been lost with a rapid rate. Therefore, monitoring a spatiotemporal distribution of mangrove is thus critical for natural resource management. This research objectives were: (i) to map the current extent of mangrove in the West and Central Africa and in the Sundarbans delta, and (ii) to identify change of mangrove using Landsat data. The data were processed through four main steps: (1) data pre-processing including atmospheric correction and image normalization, (2) image classification using supervised classification approach, (3) accuracy assessment for the classification results, and (4) change detection analysis. Validation was made by comparing the classification results with the ground reference data, which yielded satisfactory agreement with overall accuracy 84.1% and Kappa coefficient of 0.74 in the West and Central Africa and 83.0% and 0.73 in the Sundarbans, respectively. The result shows that mangrove areas have changed significantly. In the West and Central Africa, mangrove loss from 1988 to 2014 was approximately 16.9%, and only 2.5% was recovered or newly planted at the same time, while the overall change of mangrove in the Sundarbans increased approximately by 900 km 2 of total mangrove area. Mangrove declined due to deforestation, natural catastrophes deforestation and mangrove rehabilitation programs. The overall efforts in this study demonstrated the effectiveness of the proposed method used for investigating spatiotemporal changes of mangrove and the results could provide planners with invaluable quantitative information for sustainable management of mangrove ecosystems in these regions.

Estimation of Above-Ground Mangrove Biomass Using Landsat-8 Data- Derived Vegetation Indices: A Case Study in Quang Ninh Province, Vietnam

“…The classification result in this study is acceptable to compare with other previous studies (e.g. Dan et al, 2016;Tieng et al, 2019;Hai-Hoa et al, 2020b). A similar ISODATA clustering method was applied to classify mangrove forests in Philippines and the classification accuracy reached 96.6%, with a Kappa coefficient of 0.926 (Long and Giri, 2011).…”

“…Especially, the overall map accuracy was 80.0% (Table 4) with the Kappa coefficient of 0.74. This coefficient indicates that there is a substantial agreement between data collected from the field survey and the results-derived Landsat-8 image classification (Malarvizhi et al, 2016;Dan et al, 2016). During accuracy assessments, mapping accuracy might be affected by possible mixed-pixel issues, tidal regime and cloud cover percentage (Hai-Hoa et al, 2020b).…”

“…To evaluate the accuracy of Landsat-8 image classified and assess the accuracy of land use and land cover map in 2019, user's, producer's and overall accuracy with Kappa coefficient were derived from the error matrix. Kappa value is categorized into different groups where kappa value is less than zero, it then indicates no agreement; from 0 -0.20 refers to a slight agreement; 0.21 -0.40 represents as a fair agreement, from 0.41 -0.60 is considered as a moderate agreement, whereas 0.61 -0.80 is regarded as substantial, and 0.81 -1.00 refers to an almost perfect agreement (Malarvizhi et al, 2016;Dan et al, 2016). The validation GCPs were derived from the field survey in 2020; and PlanetScope and Google Earth images in 2019 (Table 2).…”

Mangrove restoration in Vietnamese Mekong Delta during 2015-2020: Achievements and challenges

“…The overall accuracy of the classified maps for 2015 and 2020 were 90.33% and 93.00%, respectively, and the kappa coefficient were 0.81 and 0.86, respectively (Table 1). These accuracy metrics showed an acceptable agreement between the classification results and reference data (Dan et al, 2016;Thomas et al, 2018;Nguyen et al, 2020b).…”

Mangrove Forest Landcover Changes in Coastal Vietnam: A Case Study from 1973 to 2020 in Thanh Hoa and Nghe An Provinces

“…The producer and user's accuracies of other land use were lower than the other classes, which could be due to spectral similarity between other land uses (such as rice fields) and other forests. However, the kappa coefficient was 0.97 (Table 4), showing that there was a substantial agreement between the classification results and reference data [83,84,115].…”