Data Paper Zone II Versions EN2 Vol 4 (2) 2019
Coastline dataset of Hainan Island during 1987–2017
: 2018 - 11 - 04
: 2018 - 11 - 23
: 2019 - 05 - 06
251 1 0
Abstract & Keywords
Abstract: Six types of coastline interpretation indicators and position extraction rules were established based on a combination of the Landsat TM//OLI data of 40 scenes (with a full coverage of Hainan Island) for eight periods during 1987– 2017 and field measurement data. The position and types of coastlines are carefully extracted with the visual interpretation and digital vectorization, by using ENVI 5.2 and ArcGIS 10.1 software. Then the coastline dataset of Hainan Island was generated with verification through field measurement data and modification. The dataset can be used to analyze the temporal and spatial variation of the coastline changes, study the characteristics of coastline exploitation and the evolution of land and sea pattern. The time-series coastline dataset could help to analyze the coastal environmental changes of Hainan Island.
Keywords: Hainan Island; coastline; coastline changes; coastal environmental changes
Dataset Profile
English titleCoastline dataset of Hainan Island during 1987–2017
Corresponding authorZhang Li (
Data author(s)Bi Jingpeng,Zhang Li,Song Xixi,Sui Yan, Wen Li
Time range1987-2017
Geographical scopeThe coastline dataset covers 12 coastal cities of Hainan Island, including Haikou, Chengmai, Lingao, Danzhou, Changjiang Li Autonomous County, Dongfang, Ledong Li Autonomous County, Sanya, Lingshui Li Autonomous County, Wanning, Qionghai, and Wenchang.
Spatial resolution30 mData volume1.75MB
Data format*.shp
Data service system
Source(s) of fundingMajor Science and Technology Program of Hainan Province (ZDKJ2016021); Major Program for Big Data Development of the National Development and Reform Commission (2016-999999-65-01-000696-01).
Dataset/Database compositionThis dataset includes coastline data products for eight time periods in Hainan Island, stored as a compressed file named “1987 – 2017 Hainan Island 8 coastline datasets.rar” that consists of eight folders named by year. It has a total amount of 1.75M. The type and length of the coastlines are recorded in the "type", length" fields of the data table respectively.
1.   Introduction
Hainan Island is an important node of the 21st Century Maritime Silk Road. The implementation of national policies such as China (Hainan) Pilot Free Trade Zone has accelerated the construction and development procedure. Coastline of Hainan Island was affected by natural and anthropogenic factors and changed severely in the last 30 years. Unique geographical location of Hainan island brings abundant coastline resources. Time series coastline monitoring has the vital significance in protecting the coastal biodiversity, resources exploitation and the protection and sustainable development of coastal ecological environment. It has great significance to monitor Hainan Island coastline change for coastal zone protection and utilization. At the same time, coastline monitoring can provide decision-making support tool for government departments to strengthen the supervision and management of coastal resources, which has great significance to the economic development, management and utilization of coastal zones. In this paper, based on multi-temporal remote sensing data, combined with field investigation records and the coastline type interpretation features, eight coastline datasets from 1987a to 2017a are obtained by the method of visual interpretation and digital vectorization, and data sharing services are provided. As a part of the coastal zone ecological environmental monitoring of Hainan Province, the datasets could reflect the spatio-temporal pattern of ecological environmental change in Hainan Island.
2.   Data collection and processing
2.1   RS Image sources
The coastline datasets were extracted by 40s least-cloud and non-gap Landsat TM/OLI images obtained from the United States Geological Survey ( ) which has 30m spatial resolution, including 1987a, 1990a, 1995a, 2000a, 2005a, 2010a, 2015a, 2017a time-series, and time-adjacent year data as supplementary. The detailed image information is shown in Table 1, and whole Hainan Island 8-periods images in 1987-2017a were obtained by the method of image mosaicking.
Table 1   Landsat images which covered Hainan Island in 8-periods
NO.Periods and Acquired dateSatelliteSensorPath-row
2.2   Data processing
Different coastline types have different interpretation and position definition methods. Establishing the position definition methods and detecting visual interpretation features are the most important aspects in coastline study. In the production process of this dataset, the coastline location and types are defined according to the geographical environment. According to the geophysical and geological marine and onshore surveys, the coastline material composition, as well as the development status of the coast, and especially the estuary coastline, are represented with the interpretation of the information about different geomorphology and setting of the coastline in Hainan Island. Some kinds of features are also be used in remote sensing images visual interpretation, including color, hue, texture, shadow, shape, patches shape, position and so on. [1 - 2]
The coastline is classified into natural coastline and artificial coastline. The natural coastline is divided into six types, e.g. estuary coastline, bedrock coastline, sandy coastline, biological coastline and silty coastline. Artificial coastline includes port construction, Arti-building coastal and other human utilized types. For example, according to the using pattern of coastline, artificial coastline types can be classified into Mari culture, salt-field, maricultural land using, coastal construction, port transportation, traffic line and protective seawall, groin etc. [3-5]
Combined with the "color", “shape”, "position" interpretation features constructed by the spectral, texture, underlying surface and geographical location of the coastline showed on the Landsat images, with the help of enhancement methods (such as image overlaying, image stretching, principal component analysis, etc.). The coastline type interpretation features were established through the optimal band combination and image enhancement in ArcGIS 10.1 software, as well as field verification of sample data. [6-7] (Table 2)
Table 2   Coastline interpretation features
Class level Ⅰ
Class level Ⅱ
Images (TM 642 false color)Coastline distribution and surfaces characters
estuary coastlineThe estuary coastline is generally located at the mouth of the river, which is the boundary line of the river and the ocean, and mostly distributed at the mouth of a river which more than 100 meters wide.
bedrock coastlineThe bedrock coastline is formed by geological tectonic activity and wave action, which consists of bare hard rock and steep terrain generally. Extensibility capes with bay associate exits often there. The coastline between headlands is tortuous, erosive and interlaced. Erosion occurs at the headland, and accumulates in the bay usually.
The sandy coastline is distributed into the open bay between the headland. The sandy coast is mainly composed of fine sand, silty sand and silt. The beach is generally wide and the coastline is usually straight and long. There are many accumulated landforms such as sand dams and offshore dams. The coast is generally in the state of erosion and retrogression, such as Sanya Bay, Yalong Bay.
biological coastlineThe biological coastline is a kind of special coast which comprises reef corals, mangroves etc. There are mangrove patches around the biological coastline, thus forms natural mangrove landscape , which has great benefit to weakening the velocity of wave movement and protecting the coastal area. Most of the biological coastline in Hainan Island is mangrove which grows, in the upper part of intertidal zone.
silty coastlineThe silty coastline is often distributed in the hidden bay which is mostly formed by muddy beach, with silt, clay and plant humus stacking, mostly displayed bluish gray or bluish black. The hydrodynamic conditions of this coastline are weak, and there are generally tidal trenches. .
(maricultural coastline)
Artificial coastline refers to the coast which original natural state changed by human activities. It is distributed in the area where there are ports, wharves, ponds, salt fields, tidal defenses, breakwaters, tidal gate, artificial Island and other buildings located.
The coastline position can be classified by a wide of sea and land boundary lines, such as the low tide line, the high tide line (mean high tide line, multi-year high tide line, etc.), the mean sea level and land boundary, and the instantaneous water boundary line, trace line or dry-wet boundary line etc. [8 - 9]. The definitions and utilization categories of coastline positions are diverse in different countries and regions, in different branches of disciplines and in different research purposes [10]. With China's relevant national standards, scholars at home and abroad mostly use the annual mean high tide line as the criterion for defining the coastline [9,11 -14]. Actually, the coastline position also affected by imaging time and tidal and ocean current conditions etc. Hence, it is difficult to accurately delineate the coastline location. Therefore, based on various techniques and methods to extract the indicated coastline, the interpretation features produce six coastline types after local tidal level correction was carried out according to the data imaging information. The coastline theoretical accurate position was obtained by modifying and relocating the coastline feature points, and coastline precision was done with evaluated indexes. In this paper, six coastline types covering Hainan Island are defined based on remote sensing image interpretation features, and their location and type of distinguish rules. [6 -7] (Table 3).
Table 3   Coastline position definitions
Class level Ⅰ
Class level Ⅱ
Coastlines shapes and position definitions
In the false color image (TM 642/ OLI 753 band combination), the coastline is the water and land boundary, the management line or the confiting line (the sealing line of the estuary) at the sudden broadening of the estuary. The estuary coastline can be extended to the boundary of the constructions, like bridges, roads and tidal gates.
In the false color image (TM 432/ OLI 543 band combination), the bedrock coastline is irregular and serrated, the rocks near the shore are gray and dark, and the plantin a red tone. There is no sediment deposit on the strongly eroded coast, and usually distributed with reefs and erosion cliffs. The weak eroded coasts have wider sediment deposits. The position is defined at the boundary between rocks and sea water.
In the true color images (TM 321/OLI 432 band combination), the sandy coastline is mostly smoother, the beach is dry with high reflectivity. The beach reflectivity is low in the submerged area, and the tone is darker. Therefore, the obvious edge of light and dark tone is the position of sandy coastline.
biological coastlineIn the false color images (TM 432/OLI 543 band combination), the mangroves show a dark red block with smooth texture and irregular shape, which is in obvious contrast with terrestrial plants. Thus, the biological coastline position should be defined in the upper edge which is submerged by the ocean tide current. In addition, if the coastline was occupied by artificial constructions such as ponds, salt lands and other artificial constructions, the type of coastline should no longer have been attributed to biological types, and would be classified into artificial coastline types.
The silty coasts are often used for fishing, and the coastline position are stable with wave dikes \ along the shore. The tidal flat of the natural silty coast has wide range. In the false color image (TM 432/OLI 543 band combination), the salt-tolerant plants grow in the land side and showing a dark red tone. The silty coastline position is defined at the land-side vegetation and the shore beach boundary.
In the true color images (TM 321 /OLI 432 band combination), there are many factories and residential areas along with the port building, most of the constructions showing grayish white, roads have obvious shape and wharf’s has regular shape outside, and the water cannot inundate its outlines. Therefore, the out-boundary of the building is artificial coastline; tidal defenses, protective seawall are man-made dykes which building on coastal beaches as well as the entrance of rivers, for preventing seawater intrusion and storm surges. the constructions outside mostly are silty beach, inside mostly are undeveloped muddy beaches or sand fill bodies. The periphery of the dam is in the location of the artificial coastline; the fishing pond and salt field is in dark blue tone; the crystal pool of salt field is in bright white tone. Also, the outer edge line of fishing pond and salt field is in artificial coastline. In addition, the outer edge has expanded the curvature and utilization length of the coastline after the formation of a certain scale, like the man-made island, the construction of reclamation. So, the outer boundary is the artificial coastline.
After obtaining the image interpretation features and defining definition standards of coastline, based on the Landsat OLI remote sensing image data of 2017a, the coastline position and type information of 2017a was extracted by automatic method firstly, and the accuracy verification, position correction and type correction were also conducted. (Figure 1). Taking the 2017a coastline data as the base data, the rest time period (1987a, 1990a, 1995a, 2000a, 2005a, 2010a,2015a) coastlines are obtained by superimposing the remote sensing images of corresponding period and applying vectorization method.
The 2017a coastline extraction method go as follows: based on the combination of remote sensing and GIS information extraction technology, firstly, the improved normalized differential water body index (MNDWI) was used to segment the pre-processed remote sensing images with global threshold, and the binarization image was generated; secondly, the gray noise patches was filled with mathematical morphology operator. Canny operator was used to extract the coastline of binary image under MATLAB platform [6]. The coastline obtained by this method is only land-water line symbols. Finally, according to the sea-charts, topographic maps and tidal data, the instantaneous line was corrected to real coastline position.

Figure 1   Coastline accuracy evaluation method
3.   Sample description
3.1   Data composes
This dataset includes a total of 8 periods of coastline location and type data of Hainan Island in the period of 1987- 2017, which are stored as a compressed file ("1987-2017a Hainan Island 8 coastline datasets .rar"). Each periods of data are stored separately in a folder named by the year, with a total amount of 1.75 MB. Each folder stores the corresponding .shp vector data file separately, and the vector data adopts WGS84-UTM coordinate projection system. Each folder with a year name contains two pieces of data, one of which contains all information attributes such as location information, length information, type information, etc. such as coastline data which named like "2010coastline_class.shp". The other contains only location information and length information, such as coastline data which named like "2010coastline.shp", which can be used for research in different fields.
3.2   Data samples
Based on the coastline data in 2017a, fixing display scale in ArcGIS software, Hainan Island coastline in 1987a, 1990a, 1995a, 2000a, 2005a,2010a and 2015a were obtained based on the interpretation features and position definition methods in Chapter 1.2. The distribution characteristics of Hainan Island coastline are shown in figure 2.









Figure 2   Coastline distribution of Hainan during 1987 to 2017
4.   Quality control and verification
4.1   Data results and accuracy verification road line
In 2017a, coastline datasets were obtained by automatic extraction methods as well as digital vectorization. With the help of DSAS (Digital Shoreline Analysis System, USGS) software, the automated extraction results are quantitatively compared with the comparing data extracted by visual interpretation, and the error deviation of coastline position was corrected. In addition, the coastline field investigation surveys in 2018 for coastline verification was also conducted (,Key Laboratory of Digital Earth Science, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences), and the coastline type of 2017a is verified by the field verified sample points (Figure 1).
The coastlines of the remaining phases (1987a, 1990a, 1995a, 2000a, 2005a, 2010a, 2015a) were generated based on 2017a coastlines. Combined with images’ imaging time, the location and type information were modified and corrected [14 - 15]. (Figure 2).
4.2   Data accuracy verification method
The coastline type interpretation and verification methods were based on field investigation, which mainly concentrated in the key areas of coastline changes, including Haikou City (such as Mangrove Nature Reserve of Dongzhai Port). Danzhou City (such as artificial island construction like Haihua island), Sanya City (such as Da dong bay, Xiao Dong bay and Coral Reef Nature Reserve in West island), Wenchang City (such as aquaculture area in Hui Wen district). The coastline types and historical changes were obtained by filed survey consulting with locals and communicating with relevant agencies.[16] (Table 4)
Table 4   Coastline survey record table
NO.LongitudeLatitudeCoastline typeTimeRecorderChangeOthers
1109°30′40″E18°31′01″NSandy coastline20180118Bi JingpengFlat tidal zone;
Straight coastline
4.3   Data precision
The results in 2017a in Wenchang area which generated by automatic coastline extraction were compared with real coastline location, and the coastline position errors are statistically analyzed by BASELINE method with DSAS software. By setting about 2 000s lines with directive information, the coastline position deviation was analyzed statistically (Figure 3). The verification results show that the absolute position error is floated in the pixel level in about 95% domain between automatic results and real coastline location. The type of estuary and silty coastline, artificial coastline (especially offshore aquaculture utilizations) were corrected by the manual.

Figure 3   Coastline position evaluation method
A total of 82 verified points with coastline type information were obtained from field investigation in Hainan Island, 380 typical coastline photos were also taken, and six types of coastline, such as biological coastline (mainly formed with mangrove and coral beef), artificial coastline, bedrock coastline and sandy coastline, were proved and verified in the field.
5.   Usage methods and advices
The coastline datasets of Hainan Island in 1987-2017a are all in shp format, and can be dealing with vector edited, attributed query, displayed by filed, and analyzed by ArcGIS and other GIS software. In the context of coastline change study of Hainan Island, most scholars choose the typical areas to analyze where the coastline change occurs greatly, but the research on the coastline change of whole Hainan Island is much less. This study analyzed the coastline changes over the last 30 years of Hainan Island in terms of coastline structure, coastline fractals, coastline change rates and coastline patterns . This paper can also be used as the basic dataset for coastal ecological environmental monitoring of Hainan Island which will help to study the temporal and spatial pattern of coastline of Hainan Island.
During field investigation, the research team had received the support and cooperation of departments and units in Hainan Province, and we are grateful for the help, and also showing our thanks to professor Liao Jingjuan , Zhu Lanwei for their suggestions on the methods of coastline type verification, master Yang Haoxiang, Bi Sen and others for their work and contribution in the acquisition of Landsat images, the discussion of automated extraction methods and technical routes and so on.
Sun Li’e. Coastline changes monitoring with remote sensing of the Zhejiang province and research on coastal vulnerability assessment[D]. Master’s Dissertation, The First Institute of Oceanography, MNR, 2013.
Sun WF, Ma Y, Zhang J, etc. Study of Remote Sensing Interpretation Keys and Extraction Technique of Different types of Shoreline[J]. Bulletin of Surveying and Mapping. 2011,3:41-44.
Liu BQ, Meng WQ, Zhao JH, etc. Variation of Coastline Resources Utilization in China from 1990 to 2013[J]. Journal of Natural Resources. 2015,30(12):2033-2044.
Yao XJ, Gao Y, Du YY, etc. Spatial and Temporal Changes of Hainan Coastline in the Past 30 Years Based on RS[J]. Journal of Natural Resources. 2013,28(1):114-125.
Zhang Y, Li XM, Ma Y, etc. Analyzing Shoreline Changes of the Pearl River Estuary Based on remote Sensing[J]. Hydrographic Surveying and Charting. 2014,34(3):52-55.
Sui Y, Zhang L, Mu XD, etc. Coastline change monitoring and analysis with remote sensing in Hainan Island[J]. Journal of Marine Sciences. 2018,36(2):36-43.
Sui Y. Monitoring and analysis of spatial-temporal changes of coastline in Hainan island in recent 30 years based on remote sensing[D]. Shandong University of Science and Technology, 2018.
Wu XJ, Xiao CC, Cui ZY, etc. Coastline Extraction Based on Object-oriented Method Using GF-2 Satellite Data[J]. Spacecraft Recovery & Remote Sensing. 2015,36(4):84-92.
Hou XY, Wu T, Wang YD, etc. Extraction and accuracy evaluation of multi-temporal coastlines of mainland China since 1940s[J]. Marine Sciences. 2014,38(11):66-73.
Yu CX, Xu J, Huang WQ, etc. Discussion on the mapping of coastline[J]. Engineering of Surveying and Mapping. 2015,24(7):1-5.
Xu JY, Zhang ZX, Zhao XL, etc. Spatial-temporal analysis of coastline changes in northern China from 2000 to 2012[J]. Acta Geographica Sinica. 2013,68(5):651-660.
Gao Y, Zhou CH, Su FZ, etc. Coastline extraction from remote sensing images based on multi-features approach[J]. Engineering of Surveying and Mapping. 2014,23(5):1-5.
Gao Y, Su FZ, Zhou CH, etc. Scale Effects of China Mainland Coastline Based on Fractal Theory[J]. Acta Geographica Sinica. 2011,66(3):331-339.
Sun WF. Research on the shoreline change of Lai Zhou bay from 1978 to 2009[D]. The First Institute of Oceanography, MNR,2010.
Zhong QQ. Coastline change monitoring and analysis with remote sensing in Ri Zhao[D]. Qu Fu Normal University. 2016.
Special Projects for Comprehensive Marine Survey and Assessment in China's Offshore Areas, State Oceanic Administration. Technical specification for coastal zone investigation[M]. Beijing: China ocean press. 2005.
Data citation
BI JP, Zhang L, Song XX, etc. Coastline dataset of Hainan Island during 1987–2017. Science Data Bank, DOI: 10.11922/sciencedb.669 (2019).
Article and author information
How to cite this article
BI JP, Zhang L, Song XX, etc. Coastline dataset of Hainan Island during 1987–2017. China Scientific Data 4(2019). DOI: 10.11922/csdata.2018.0066.zh
Bi Jingpeng
Major contribution: Landsat TM/OLI data acquisition, coastline types interpretation and automatic extraction method study, etc.
Master student; research area: Remote Sensing information extraction and applications.
Zhang Li
Major contribution: quality and technical design in coastline dataset producing.
PhD, Professor, research area: Remote sensing applications in global ecology and coast science.
Song Xixi
Major contribution: dataset accuracy verification and evaluation, field verification and investigation in Hainan island.
Master student; research area: Remote sensing applications in coastal zone.
Sui Yan
Major contributions: coastline types interpretation and information extraction, accuracy verification and evaluation, etc.
Master student; research area: Coastline time-series changes and coastal development situation study.
Wen Li
Major contributions: coastline types interpretation and extraction methods designation.
Senior engineer; research area: Remote sensing technology applications in land investigation filed.
Major Science and Technology Program of Hainan Province (ZDKJ2016021); Major Program for Big Data Development of the National Development and Reform Commission (2016-999999-65-01-000696-01).
Publication records
Published: May 6, 2019 ( VersionsEN2
Released: Nov. 23, 2018 ( VersionsZH2
Published: May 6, 2019 ( VersionsZH4
Updated: May 6, 2019 ( VersionsZH5