Abstract: Distributed in the coastal wetlands of tropical and subtropical regions throughout the world, mangrove forests provide important ecological and eco-economical services such as coastal erosion prevention, water purification, nursery habitats for marine fish and shrimp breeding, provision of building materials and medicinal ingredients, and tourist attraction. A dataset of mangrove forests changes from 1987 to 2017 in Hainan Island was developed using Landsat TM/OLI images obtained in 1987, 1993, 1998, 2003, 2007, 2013 and 2017. This dataset was compiled through a support vector machine (SVM) classification method based on ground survey data. It can be used to understand the status and dynamics of mangrove forests, and to support decision-making concerning the restoration, protection and management of mangrove wetland ecosystem. Furthermore, the dataset provides basic data for ecological environment monitoring in Hainan Province.
Keywords: Hainan Island; mangrove forests; Landsat image; change monitoring
|Title||A dataset of mangrove forests changes in Hainan Island during 1987–2017 based on Landsat data|
|Data corresponding author||Liao Jingjuan (firstname.lastname@example.org)|
|Data authors||Liao Jingjuan, Zhen Jianing|
|Geographical scope||Hainan Island|
|Data volume||4.12 MB|
|Data format||*.dbf, *.prj, *.shp, *.shx|
|Data service system||http://www.sciencedb.cn/dataSet/handle/677|
|Sources of funding||Major Science and Technology Program of Hainan Province (Grant No. ZDKJ2016021); Strategic Priority Research Program of the Chinese Academy of Sciences (A-level; XDA19030302).|
|Dataset composition||This dataset consists of seven subsets corresponding to seven different mangrove distribution sites in Hainan Island, including Dongzhaigang, Qinglangang, Maliaogang, Huachangwan, Xinyinggang, Yangpugang and Dongfang. Each subset includes mangrove area data of seven phases extracted from Landsat images during 1987-2017. Data of each phase includes four types of files, namely, DBF, PRJ, SHP, and SHX.|
Mangrove forests are distributed in the coastal wetlands of tropical and subtropical regions throughout the world. They are swampy woody plant communities, which consist of evergreen shrubs or trees, and have special sea-land characteristic and enormous ecological, economic and social values1.2.3. . They play an irreplaceable role in maintaining biodiversity, protecting coastal environment, strengthening dyke and against from wind, protecting bank and inducing siltation, purifying the coastal water environment and protecting farmland and village from some natural disasters such as hurricanes and tsunamis184.108.40.206.8.9. . At the same time, the forests belong to the most threatened and vulnerable ecosystems, and are under threat from both natural and anthropogenic forcing. For nearly half a century, human continue to invade and cut down mangroves, due to the rapid development of social economy and coastal economic zones. Owing to changes in hydrological conditions, climate change and water pollution, the world mangrove forests are declining at an alarming rate and lost 36% between 1995 and 2005, perhaps even more rapidly than inland forests and tropical rainforests, and much of which remains is in degraded conditio10.11.12.13.14. .
In China, mangrove forests are mainly distributed in the southern and southeastern coastal zones of mainland, as well as in the coastal zones of Hainan Island and western Taiwan, with a coastline length of approximately 14 000 km. Hainan Island has high mangrove species-richness and a wide distribution of mangrove forests. The mangrove forests are distributed in the northeast, south, west and east of the island, and contain the vast majority of mangrove species in China. The mangrove communities are complex and belong to the typical oriental group.
The mangrove forests are distributed in harsh environmental settings such as high salinity, high temperature, extreme tides, high sedimentation and muddy anaerobic soils. So, it is difficult to obtain accurate data by the field investigation. Many studies have shown that remote sensing is the tool for the provision of distribution and dynamic changes quickly and accurately220.127.116.11.19.20. . These researches focus on the global or national scale of mangrove forests mapping. Most of the obtained datasets are a specific year or a certain period of time. Two global scale mangrove forest maps were produced for the year 2000 by Spalding et al. and Giri et al.21.22. ， and Hamilton. and Casey released global scale mangrove forest from 2000 to 201223.. Recently, Chen et al. mapped mangrove forests of China in 2015 using time series Landsat and Sentinel-1A images24.. In recent years, the mangrove forests in Hainan Island have a large change due to rapid economic development and the impacts of human activity. However, this area is lack of long time series mangrove forests dynamic changes dataset. So, a dataset of mangrove forests changes from 1987 to 2017 in Hainan Island was developed using multi-temporal Landsat TM/OLI images. This dataset was compiled through a support vector machine (SVM) classification method based on ground survey data. It can be used to support decision-making concerning the restoration, protection and management of mangrove wetland ecosystem.
2.1 Data acquisition
In this study, multi-temporal Landsat images were selected to monitor long-term changes in Hainan Island mangrove forests. The Landsat images obtained between 1987 and 2017 were downloaded from the USGS Center for Earth Resources Observation and Science (http://www.usgs.gov). A total of 39 Landsat images were downloaded, including 28 Landsat Thematic Mapper (TM) from 1987, 1993, 1998, 2003 and 2007 respectively, and 11 Landsat Operational Land Imager (OLI) images. A complete list of the Landsat images used in this study is shown in Table 1.
Table 1. Landsat images used in this study
At the same time, in order to validate the dataset obtained from the multi-temporal Landsat images, the Chinese Gaofen-2 (GF-2) images obtained in 2015 were also used in this study. A total of 15 GF-2 images were downloaded from the Chinese High-Resolution Earth Observation System-Hainan Data and Application Center, including two high-resolution cameras (PAN images with a resolution of 1 m and MS images with a resolution of 4 m). Additionally, the field survey was carried out during the period of December 2016, March 2017 and January 2018 respectively. The survey collected training and validation data and then invest mangrove forests species, distribution, growth condition and environment. We selected 386 ground truth points, including 152 points of mangroves and 234 points of other land cover types. These ground truth points contained spatial location and land cover types, and were used to validate the range and classification of mangrove forests from Landsat images.
2.2 Data processing
The Landsat images were already geo-referenced, so the ENVI software was used to perform the radiometric calibration and atmospheric corrections on the Landsat images. The FLAASH module in ENVI software was used for atmospheric corrections of Landsat images. In order to standardize the dataset, we rectified the Landsat images obtained in 1987, 1993, 1998, 2003, 2007 and 2013 using the Landsat OLI images obtained in 2017 as a master dataset. An average root mean square error (RMSE) of less than 0.5 pixels was obtained for the co-registered images. The GF-2 images were performed the orthorectification, radiometric calibration, geometric rectification, and image fusion. The subtractive resolution merge method in ERDAS IMAGE software was used to fuse PAN and MS images of GF-2, and to obtain the fusion images with a resolution of 1m.
In this study, using the fusion images of GF-2 with a resolution of 1m, we subset the areas of mangrove reserves from the GF-2 images based on the boundaries of mangrove reserves. A support vector machine (SVM) classification method in ENVI software was used to perform the land cover types classification in Hainan mangrove forest reserves. To obtain the optimal results of Hainan mangrove forests in 2015 from GF-2 images, the visual interpretation was performed to confirm the extraction of mangroves based on the field data and Google Earth images, which means the manual modification of misclassified objects. Then, the SVM classifier was used to classify the Landsat OLI images obtained in 2017, and the best classification results were obtained by the manual modification based on the classification results of GF-2 and Google Earth images in 2015. Finally, the mangroves and other land cover types in the Landsat images obtained in 1987, 1993, 1998, 2003, 2007 and 2013 were extracted based on the classification results of Landsat OLI images obtained in 2017.
3.1 Dataset information
A dataset of mangrove forests changes from 1987 to 2017 in Hainan Island was developed using Landsat TM/OLI images obtained in 1987, 1993, 1998, 2003, 2007, 2013 and 2017. The dataset contains the changes of mangrove forests in Dongzhaigang National Nature Reserve, Qinglangang Provincial Nature Reserve, Huachangwan, Maliaogang, Xinyinggang, Yangpugang and Dongfang, and the distribution of each site is shown in Figure 1.
The dataset consists of 7 folders. Each folder is named after the above 7 sites. Qinglangang is composed of three sections (Huiwen, Puqian, Guannan), and includes three sub-folders. The other 6 sites consist of six folders, each folder contains the result of 7 years mangrove extraction. The file format is shp and the naming rule is “place name+year_mangrove.shp. The details of the dataset are listed in Table 2.
Table 2. The information of Hainan Island mangrove changes datasetduring 1987–2017
3.2 Data samples
The datasets can be used to generate the distribution and changes of mangrove area in ArcGIS software. The distribution of mangrove area in Dongzhaigang in 1987 and the changes of mangrove area from 1987 to 2017 are shown in Figure 2 and Figure 3.
In this study, Kappa coefficient and the confusion matrix are used for accuracy assessment of the classified images. First, we utilized Biogeography Branch’s Sampling Design Tool in ArcGIS to generate sampling points in 7 key mangrove distribution areas randomly. Then, circular buffers with a radius of 9m are created according to these points. Based on these circular buffers, circumscribed rectangles were built with the help of ArcGIS. The rectangles are then divided into mangroves and non-mangroves according to Google Earth images and field verification points. At the same time, we added some polygons in some sparse areas. Finally, these polygons were used to validate the accuracy of mangrove classification. The classification of Landsat in 2017 and GF-2 images were evaluated. The number of GF-2 verification points is 607, including 354 mangrove points and 253 non-mangrove points. The total accuracy of classification is 99.0%, with Kappa coefficient is 0.98.
Based on the result of GF-2 images classification, 300 mangrove verification polygons were generated randomly. Then another 350 non-mangrove verification polygons were generated randomly according to Reserve (except mangrove region). The Landsat images classification in 2017 was evaluated. The total accuracy of classification is 98.8%, with Kappa coefficient is 0.98. The specific evaluation results can be seen in List 3.
Table 3. The accuracy assessment of Landsat in 2017 and GF-2 images classification
OA：Overall Classification Accuracy；PA：Production Accuracy；UA：User Accuracy；MF：Mangrove Forest；N-MF：Non-Mangrove Forest
This dataset is a relatively complete product of Hainan mangrove forests changes in nearly 30 years. It demonstrates the mangrove information of Hainan mangrove Reserves. The dataset can be used in common GIS software, such as ArcGIS, MapGIS and MapInfo. Other exchange formats can also be used. Through the statistics from different periods, the changes of mangrove area from 1987 to 2017 can be generated. Spatial distribution of mangrove changes in different regions and different periods can also be obtained according to spatial overlay analysis. This dataset provides basic data for ecological environment monitoring in Hainan Province and scientific research. It can also be used to understand the status and dynamics of mangrove forests, quality assessment of ecological environment, and to support decision-making concerning the restoration, protection and management of mangrove wetland ecosystem.
We would like to thank to the relevant departments and units in Hainan Province for supporting and coordinating the filed surveys. Thanks to the United States Geological Survey’s Earth Resources Observation and Science Center (USGS/EROS) for providing Landsat data.
Zhang Q M, Zhang Y C. Study on Biogeomorphologic Process of Mangrove Coasts in South China [J]. Quat. Sci.1997, 17, 344–353. (In Chinese)
GIRI C. Observation and monitoring of mangrove forests using remote sensing: opportunities and challenges[J]. Remote Sensing, 2016, 8(9): 783.
LUGO A E, SNEDAKER S C. The ecology of mangroves[J]. Annual Review of Ecology & Systematics, 1974, 5: 39-64.
Lin P. A Review on the Mangrove Research in China [J]. J. Xiamen Univ. (Nat. Sci.) 2001, 2, 592–603. (In Chinese)
GLASER M. Interrelations between mangrove ecosystem, local economy and social sustainability in Caeté Estuary, North Brazil[J]. Wetlands and Management, 2003, 11(4): 265-272.
MUMBY P J. Connectivity of reef fish between mangroves and coral reefs: Algorithms for the design of reserves at seascape scales[J]. Biological Conservation, 2006, 128(2): 215-222.
ALONGI D M. Mangrove forests: resilience, protection from tsunamis, and responses to global climate change[J]. Estuarine Coastal Shelf Science, 2008, 76(1): 1-13.
DONATO D C, KAUFFMAN J B, MURDIYARSO D, et al. Mangroves among the most carbon-rich forests in tropics[J]. Nature Geoscience, 2011, 4(5): 293-297.
BLASCO F, AIZPURU M, GERS C. Depletion of the mangroves of Continental Asia[J]. Wetlands Ecology & Management, 2001, 9(3):255-266.
VALIELA I. Mangrove forests: One of the world's threatened major tropical environments[J]. BioScience, 2001, 51(10): 807-815.
ALONGI, DANIEL M. Present state and future of the world's mangrove forests[J]. Environmental Conservation, 2002, 29(3): 331-349.
DUKE N C, MEYNECKE J O, DITTMANN S, et al. A world without mangroves? [J]. Science, 2007, 317(5834): 41.
GIRI C, ZHU Z, TIESZEN L L, et al. Mangrove forest distributions and dynamics (1975-2005) of the tsunami‐affected region of Asia[J]. Journal of Biogeography, 2008, 35(3): 519-528.
ABDEL-HAMID A, DUBOVYK O, ELMAGD I A, et al. Mapping mangroves extents on the Red Sea coastline in Egypt using polarimetric SAR and high resolution optical remote sensing data[J]. Sustainability, 2018, 10(3): 646.
GIRI C, OCHIENG E, TIESZEN L L, et al. Status and distribution of mangrove forests of the world using earth observation satellite data[J]. Global Ecology & Biogeography, 2011, 20(1): 154-159.
GIRI C, LONG J, ABBAS S, et al. Distribution and dynamics of mangrove forests of South Asia[J]. Journal of Environmental Management, 2015, 148: 101-111.
KIRUI K B, KAIRO J G, BOSIRE J, et al. Mapping of mangrove forest land cover change the Kenya coastline using Landsat imagery[J]. Ocean & Coastal Management, 2013, 83(10): 19-24.
LONG J, NAPTON D, GIRI C, et al. A mapping and monitoring assessment of the Philippines' mangrove forests from 1990 to 2010[J]. Journal of Coastal Research, 2013, 30(2): 260-271.
RAKOTOMAVO A, FROMARD F. Dynamics of mangrove forests in the Mangoky River delta, Madagascar, under the influence natural and human factors[J]. Forest Ecology & Management, 2010, 259(6): 1161-1169.
GIRI C, OCHIENG E, TIESZEN L L, et al. Status and distribution of mangrove forests of the world using earth observation satellite data[J]. Global Ecology Biogeography, 2011, 20: 154–159.
HAMILTON S E, CASEY D. Creation of a high spatio-temporal resolution global database of continuous mangrove forest cover for the 21 st century (CGMFC-21)[J]. Global Ecology Biogeography, 2016, 25: 729-738.
Liao J and Zhen J. A dataset of mangrove forests changes in Hainan Island during 1987–2017 based on Landsat data. Science Data Bank, DOI: 10.11922/sciencedb.677(2018).
How to cite this article
Liao J and Zhen J. A dataset of mangrove forests changes in Hainan Island during 1987–2017 based on Landsat data. China Scientific Data 4(2019). DOI: 10.11922/csdata.2018.0072.zh.