Next Article in Journal
An Enhanced Residual Feature Fusion Network Integrated with a Terrain Weight Module for Digital Elevation Model Super-Resolution
Previous Article in Journal
Mapping Diverse Paddy Rice Cropping Patterns in South China Using Harmonized Landsat and Sentinel-2 Data
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Remote Sensing
Volume 15
Issue 4
10.3390/rs15041035
remotesensing-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Spatial–Temporal Evolution Monitoring and Ecological Risk Assessment of Coastal Wetlands on Hainan Island, China

by
Haiyan Chen
1,2,*,
Dalong Li
1,2,
Yaning Chen
3 and
Zhizhong Zhao
1,2
1
College of Geography and Environmental Science, Hainan Normal University, Haikou 571158, China
2
Key Laboratory of Earth Surface Processes and Environmental Change of Tropical Islands, Haikou 571158, China
3
State Key Laboratory of Desert and Oasis Ecology, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, China
*
Author to whom correspondence should be addressed.
Remote Sens. 2023, 15(4), 1035; https://doi.org/10.3390/rs15041035
Submission received: 8 January 2023 / Revised: 8 February 2023 / Accepted: 12 February 2023 / Published: 14 February 2023
Browse Figures
Review Reports Versions Notes

Abstract

:
Coastal wetlands are located at the intersection of land and sea and provide extremely important ecological services. The coastal wetlands of estuarine harbors are representative parts of the coastal wetlands. Changes that occur in estuarine harbor wetlands are microcosms of the changes occurring in the coastal wetlands more generally. The coastal wetlands of Hainan Island, China, are coastal wetlands typical of tropical islands and are extremely sensitive to climate change. In the context of global sea level rise, studying the characteristics of spatial and temporal distribution of coastal wetlands on Hainan Island, as well as changes in their vulnerability, could provide scientific and technological support to address the adverse effects of climate change. Using nine typical estuarine harbor wetlands as target areas, this study systematically studies the spatial–temporal evolution of coastal wetlands on Hainan Island from 1990 to 2020. The results suggest the following: (1) The total area of coastal wetlands has remained relatively stable, but the area of artificial wetlands, especially aquaculture ponds, has increased significantly. There is a clear spatial variability in the changes in mangrove wetlands, with a clear increase in the area of areas with a high degree of protection, such as Dongzhai Harbor (DZG). The area of the areas with a high intensity of human activity has been significantly reduced, such as Bamen Bay (BMG). (2) The overall ecological risk of coastal wetlands is low, with the average wetland risk index (WRI) of all harbors being below 0.15. The higher the degree of protection, the lower the ecological risk of the area, such as DZG. Human activities are the main factor causing increased ecological risk in wetlands. (3) Climate-change-induced sea level rise and the intensification of human activities are the main determinants of future trends in the spatial distribution of coastal wetlands and wetland ecosystem stability. The results of this study provide guidance on the conservation and restoration of coastal wetlands.

1. Introduction

Located at the intersection of land and sea, not only do coastal wetlands perform the functions of coastal protection, water purification, biodiversity protection, and the promotion of tourism and recreational value, but their function as a carbon sink is much higher than that of other ecosystems [1,2,3]. The rate of carbon burial of the blue carbon systems of coastal wetlands is tens to thousands of times higher than that of terrestrial forest systems per unit area, and about 50 times higher than that of marine ecosystems [4,5,6]. In the context of global warming, coastal wetland blue carbon ecosystems will play an important role in dealing with climate change [5,7]. Despite these functions, coastal wetlands are threatened by climate change and human activities [8,9,10,11]. Almost 4000 km2 of coastal wetland have been lost in the last 20 years [12].
The study of spatial–temporal changes in coastal wetlands is helpful to clarify the evolution process and trend of coastal wetlands, analyze the fragility and stability of coastal wetland ecosystems, elucidate the change process and trend of wetland’s ecological functions, and reveal the driving factors of the spatial–temporal evolution of coastal wetlands. This will provide scientific support for the protection and restoration of coastal wetlands, promote the capacity of coastal wetlands to sequester carbon, and improve the coastal wetlands’ ability to cope with climate change.
Based on remote sensing data, the study of spatial–temporal changes in coastal wetlands has made remarkable achievements [13,14]. On a global scale, Murray et al. [8] found a global loss of 16.02% of coastal wetlands between 1984 and 2016 based on Landsat data. From 1999 to 2019, the net loss area of global coastal wetlands reached 4000 square kilometers, and Asia was the region with the largest loss [12]. A large number of scholars have also studied the spatial–temporal variation of coastal wetlands at the regional scale based on different remote sensing data. For example, based on Landsat data or the combination of Landsat data and other remote sensing data, some scholars studied the spatial–temporal changes of coastal wetlands in a long time series (more than 25 years) in Northeastern Italy (1984–2016) [15]; the Mekong Delta in Vietnam (1995–2020) [16]; the southeast part of the Alligator River National Wildlife Refuge in the USA (1995–2019) [17]; the coast of southeast China (1990–2015, 1973–2015) [18,19]; on the border of the Yellow Sea (1978–2018) [20]; and in the Yancheng of China (1991–2021) [21], Jiaozhou Bay of China (1980–2020) [22], and Dongzhai Harbor of China (1950–2020) [23]. Some scholars have also taken typical coastal wetlands as examples and combined multisource, high-resolution remote sensing data to propose a coastal wetland classification method with higher accuracy, providing new methods to improve coastal wetland monitoring [24,25,26,27,28]. Aslan et al. mapped the spatial distribution of mangrove composition, canopy height, and aboveground biomass in Mimika, Papua, Indonesia, in 2013 based on a classification method that combines active and passive remote sensing data [24]. Zhang et al. proposed a coastal wetland classification method based on a gravitationally optimized multilayer perceptron and morphological attribute profiles based on Sentinel-2 data and verified the effectiveness of the method in the Jiaozhou Bay and the Yellow River Delta [25]. Chang et al. proposed a method for detecting sub-pixel changes in coastal wetlands based on collaborative coupled hyperspectral unmixing, and the accuracy of the method was verified in the Yellow River Delta based on Gaofen-5 and FY-1 02D hyperspectral images [26]. Liu et al. detected changes in coastal wetlands in the Yellow River Delta during 2014–2019 based on multiband SAR coherence and synergetic classification [27]. Yuan et al. mapped the distribution of the coastal wetlands in the Yancheng City, the Yellow River Delta, and the Hangzhou Bay based on the multiresolution synergistic coupling of SAR, multispectral, and high-resolution images [28]. However, these studies mainly focus on the spatial–temporal variation in coastal wetlands along the mainland and typical harbor wetlands. Up to now, we have poor knowledge of the spatial–temporal evolutionary characteristics of coastal wetlands on a typical tropical mountain island, which are extremely sensitive and vulnerable to climate change.
In coastal wetland vulnerability studies, a large number of scholars have studied the effects of different magnitudes of sea level rise on the stability and ecological functions of coastal wetlands under different emission scenarios in the future [23,29,30,31,32,33,34]. Schuerch et al. predicted that the total area of global coastal wetlands can be increased by up to 60% if enough space for coastal wetlands to retreat is available; otherwise, it may be reduced by 30% [29]. Saintilan et al. predicted that mangroves are likely (>90% probability) to fail to grow sustainably when the relative sea level rise rate exceeds 6.1 mm per year. Moreover, in the high-emissions scenario, the tropical coastline is likely to exceed this threshold within 30 years [30]. Estrela-Segrelles et al. evaluated the risk of coastal wetlands in the Mediterranean under different emission scenarios [31]. Cai et al. predicted that under RCP 4.5 and RCP 8.5, sea level rise is expected to result in the loss of 26% of the mangroves in Dongzhai Harbor by 2100, and under RCP 2.6, 17% of the mangroves may be lost [23]. Chatting et al. predicted the changes in mangrove carbon stocks under future climate change and deforestation [32]. By integrating data from 166 estuaries around the United States, Osland et al. predicted that the migration of coastal wetlands to land will alter coastlines, but not offset the loss to the ocean. Two-thirds of the potential wetland migration is expected to be at the expense of coastal freshwater wetlands, while the remaining one-third is expected to be at the expense of valuable uplands, including agricultural lands, forests, pastures, and grasslands [33]. Van der Stocken et al. predicted that global changes in seawater density would disrupt the spread of mangroves [34]. However, only a few scholars have attempted to assess the historical evolution of coastal wetland stability or ecological vulnerability [35,36]. Inadequate knowledge of the history of changes in the vulnerability of coastal wetlands will be an obstacle to accurately predicting their future changes.
Hainan Island is a typical mountainous tropical island, influenced by both the mainland and the ocean. Hainan Island is an important coastal wetland distribution area in China and a typical representative of wetland resources in the world. There are abundant wetland resources and a wide variety of wetlands here. Understanding the past is the basis for predicting the future. However, so far, very little information is known about the spatial and temporal evolutionary characteristics, stability, and ecological risks of coastal wetlands on Hainan Island. Therefore, this study aimed to (1) analyze the spatial–temporal evolutionary characteristics of different types of coastal wetlands; (2) assess the stability and ecological risk of each typical wetland; and (3) reveal the causes and ecological impacts of wetland evolution on Hainan Island.

2. Materials and Methods

2.1. Study Area

Hainan Island is located at latitude 18.21–20.21°N and longitude 108.67–111.27°E (Figure 1). It is in the outer tropical region of the northern part of the South China Sea, and is the largest tropical island in China. Hainan Island is bordered by the Qiongzhou Strait in the north and the South China Sea in the south, with a land area of 35,400 km2. Hainan Island has an East Asian monsoon climate. The dry season is from November to April and the rainy season is from May to October. The total annual precipitation is 1500–2000 mm, of which 35–70% is related to typhoons. The annual mean temperature is 24.0 °C. The topography of Hainan Island is characterized by mountains in the middle, surrounded by coastal plains, with elevation decreasing from the middle to the surroundings. There are many rivers and harbors in the developed mountainous water system. The coastal wetlands of Hainan Island are diverse and widely distributed, and have the highest diversity of mangrove species in China. Among the 27 species of true mangroves in China, 26 species are found on Hainan Island [37].
In this study, nine different types of estuarine harbors with different intensities of protection and human activities were selected as target areas for the study. They are Dongzhai Harbor (DZG), Bamen Bay (BMG), Changqi River Bay (CQH), Boao Harbor (BAG), Changhua Harbor (CHG), Danzhou Bay (DZW), Wenke River Bay (WKH), Diaolou Harbor (DLG), and Meilang Harbor (MLG) (Figure 1). Among them, DZG is the first mangrove reserve established in China. A provincial nature reserve has also been built in BMG. CHG is an important fishing port in Hainan. DZW is not only an estuarine harbor wetland, but also a typical inland sea.

2.2. Datasets

The datasets include remotely sensed data, observation data, etc. The remote sensing datasets include Landsat-5 TM (Thematic Mapper), Landsat-7 ETM + (Enhanced Thematic Mapper Plus), and Landsat-8 OLI (Operational Land Image) images. Data from 1990, 1995, 2000, 2005, 2010, 2015, and 2020 were selected for the study, respectively. Images that are cloud-free, from the same season, and clear were selected. If the image could not meet these requirements, an image from a previous or later year’s data was chosen instead. Since the dry season has more sunny days and less clouds than the rainy season, most of the rainy season data are not up to standard; thus, to ensure the consistency of the data, all the remote sensing data are from the dry season. Data preprocessing, such as atmospheric calibration, Landsat-7 ETM+ image strip repair, and the export of the cropped data to TIFF format, were conducted on the Google Earth Engine (GEE) platform.
In addition to spectral bands, NDVI (Normalized Difference Vegetation Index) [38], NDWI (Normalized Difference Water Index) [39], MNDWI (Modified Normalized Difference Water Index) [40], and CMRI (Combined Mangrove Recognition Index) [16,41] data were also calculated and integrated into the classification algorithms (Table 1).
Meteorological data were downloaded from the China Meteorological Data Network (https://data.cma.cn/ (accessed on 16 October 2022)). Water quality data were obtained from the China Ecological and Environmental Status Bulletin (https://www.mee.gov.cn/hjzl/ (accessed on 16 October 2022)) and the Hainan Province Ecological and Environmental Situation Bulletin (http://hnsthb.hainan.gov.cn/ (accessed on 16 October 2022)).

2.3. Classification

Our wetland classification system is shown in Table 2, based on Ramsar wetlands classification, landscape composition, and wetland classification schemes of previous studies [42].
The classification uses a combination of object-oriented classification method and visual interpretation: object-oriented classification to implement systematic classification, and visual interpretation modification to improve classification accuracy. In this study, eCognition Developer 8.64 was used to carry out the object-oriented classification. The specific classification steps are as follows.
(1) Multiscale segmentation of images. Image segmentation is one of the most important steps in object-oriented classification. This study used the plug-in of the eCognition software named estimation of scale parameter (ESP2), which relies on the potential local variance to detect scale shifts in geospatial data [43,44]. The tool detects the number of layers added to the project and iteratively segments them using a bottom-up multiresolution segmentation algorithm. The scale factor of its segmentation, i.e., the scale parameter, increases with the increment in the constant. The average local variance value of the objects is calculated in all layers as a condition to stop the iteration. When a scale level records a potential local variance value equal to or less than the previous value, the iteration ends and the object of the previous level segment is retained (Figure 2). Three orders of magnitude of scale parameter lags produce the corresponding number of scale levels.
(2) Classification. The maximum likelihood algorithm is considered to be one of the most effective methods for land cover classification using remote sensing data with medium spatial resolution [13,14]. The method assigns each segmented object to a different class based on class signatures’ means and variances. Therefore, training samples representing feature types are needed. The training samples were selected based on visual interpretation and Google Earth images (Figure 3). The separability of the spectral features of the generated training samples is evaluated to minimize misclassification between classes. When the separability is satisfied, supervised classification is performed in eCognition Developer 8.64 using the maximum likelihood algorithm. Finally, high-resolution Google Earth images and visual interpretation were used to visually modify the classification results to improve classification accuracy.
(3) Accuracy assessment. Classification accuracy was evaluated using a random test sample. Test samples were taken from historical Google Earth images. In this study, the error matrix method was used [45]. Overall accuracy, producer accuracy and consumer/user accuracy, and Cohen kappa coefficient values were generated and reported for each generated land cover map. In addition to the statically generated accuracy assessment report, we overlaid the classification results with the Landsat images of each region to visually assess the synchronization between them.

2.4. Risk Assessment

The Wetland Risk Assessment System is used to assess the disturbance of coastal wetlands by external pressures [36,46]. The assessment framework includes both external hazard indices (EHI) and internal vulnerability indices (IVI). The EHIs include natural hazards and human activities. The IVIs are calculated based on wetland area and wetland structure.
Natural disasters are mainly determined based on annual mean temperature and precipitation. The annual mean temperature and precipitation were obtained by interpolation of the temperature and precipitation data from seven meteorological stations (Haikou, Dongfang, Danzhou, Qiongzhong, Qionghai, Sanya, and Lingshui) on Hainan Island using the thin-plate sample interpolation method.
Anthropogenic impacts are mainly determined based on water quality and aquaculture pond expansion. The water quality is divided into I, II, III, IV, and V categories, representing the water quality from good to poor. The ratio of aquaculture pond area to the total area of each study area represents the impact of aquaculture pond area expansion on coastal wetlands. The land use data are derived from the above classification results.
Internal vulnerability is calculated based on wetland area and wetland structure. The ratio of wetland area (natural wetlands in the above classification results) to the area of each study area was used to assess the vulnerability of wetlands to natural hazards and human activities. Wetland structure was mainly calculated for landscape fragmentation index and aggregation index. In this study, the landscape fragmentation index and aggregation index were calculated using FRAGSTATS 4.2 (University of Massachusetts, Amherst, MA, USA) (Table 3).
To overcome the inconsistencies between different index units, each parameter is normalized as follows:
P i j = M i j − M i n M j M a x M j − M i n M j
R i j = M a x M j − M i j M a x M j − M i n M j
where P i j and R i j are the results after the normalization of each parameter. P i j is a positive value and is the parameter with a positive impact on the coastal wetlands. R i j is a negative value and is the parameter with a negative impact on coastal wetlands. M i j is the j-th parameter in the i-th study region; and M a x M j and M i n M j are the maximum and minimum values of the j-th parameter, respectively.
After normalizing each parameter, the values of EHI, IVI, and wetland risk indices (WRI) were all in the range of 0–1. Higher values indicate greater risk or vulnerability. The EHI is calculated as in Equation (3):
E H I = 1 2 × ∑ i = 1 n N D i × W i + 1 2 × ∑ j = 1 n H M j × W j
where N D i and H M j are the i-th natural hazard index and the j-th human activity index after normalization, respectively, and W i and W j are the weights of the i-th natural hazard index and the j-th human activity index, respectively. In this study, the same weight (1/2) was assigned to both natural hazard indices and both human activity indices.
The IVI is calculated as in Equation (4):
I V I = 1 2 × A a × W a + 1 2 × ∑ i = 1 n S i × W i
where A a represents the normalized wetland area parameter; S i is the i-th wetland structure parameter after normalization; and W a and W i are the weights of wetland area and wetland structure parameters, respectively. In this study, a weight of 1 was assigned to the wetland area. The two wetland structural parameters (landscape fragmentation index and aggregation index) were given the same weight (1/2).
The WRI is calculated based on the EHI and the IVI (Equation (5)):
W R I = E H I × I V I
In ArcGIS 10.6 software, we classified WRI below 0.1 as low risk level, WRI between 0.1 and 0.2 as medium risk level, and WRI above 0.2 as high risk level according to the principle of equal-variance classification [36,46].

3. Results

3.1. Accuracy Assessment

Table 4 shows the classification accuracy assessment results. In this study, the overall precision of the classification results for all years and all regions was greater than 90%, and the kappa coefficients were greater than 0.85.

3.2. Wetland Changes

Figure 4 shows the spatial–temporal variation in typical harbor coastal wetlands on Hainan Island between 1990 and 2020. DZG has the largest mangrove area, natural wetland area, and total wetland area, followed by BMG and DZW. BMG has the largest artificial wetland area, followed by DZG and DZW. The artificial wetlands in all harbors showed a clear and continuous increasing trend, indicating that the intensity of human activities on Hainan Island continued to intensify during the study period. The total area of mangrove wetlands and natural wetlands in DZG exhibited a consistent increase, but fluctuating changes were shown in other areas. However, the variation in total wetland area was small. It is noteworthy that the mangrove wetlands in BMG and CQH showed a decreasing trend until 2000 and an increasing trend from 2000 to 2010, followed by another decreasing trend. In addition, the natural wetland area of these two harbors showed an overall decreasing trend between 1990 and 2020.
Figure S1 shows the spatial–temporal variation in the wetlands in each harbor. Both natural and artificial wetlands in DZG show a gradual expansion to the south. The area of mangrove wetland patches gradually increases, and the connectivity between patches gradually increases. The natural wetlands in both the east and west of BMG are gradually being encroached upon by artificial wetlands. The mangrove wetlands and tidal flats in the east, especially, have almost all been turned into aquaculture ponds. The spatial distribution of mangroves in CQH has not changed substantially; however, the distribution of aquaculture ponds has gradually increased to cover almost the entire area over 30 years. There are no mangrove wetlands in CHG, but there are large areas of tidal wetlands; however, since 2005, large aquaculture ponds began to gradually occupy the tidal flats. A common feature of the spatial–temporal variation in other harbor wetlands is that the expansion of mangrove wetlands into tidal wetlands is limited by the gradual expansion of aquaculture ponds. Mangrove wetland patches are gradually fragmented and marginalized.
Figure 5 shows the variation in the ratio of wetland area to the area of each study area. DZG has the highest mangrove area ratio, which is generally maintained at more than 15%. This ratio showed an increasing trend (from 15.12% in 1990 to 21.89% in 2020). This increasing trend is especially significant since 2015. It shows that the earlier the protection, the better the development of natural coastal wetlands. In contrast, the mangrove area ratio of BMG decreased significantly, from 15.77% to 8.72%, indicating that human activities are an important cause of the degradation of natural coastal wetlands such as mangroves. The ratio of mangrove area in other harbors is below 10% and without a significant trend. The area ratio of artificial wetlands in all harbors showed an increasing trend. Among them, the highest artificial wetland area ratios were found in CQH (from 2.13% in 1990 to 32.50% in 2020) and BMG (from 6.06% in 1990 to 25.98% in 2020). The ratio of natural wetland area to all wetlands in each harbor is above 40%, and also has no significant trend.

3.3. Risk Assessment

Figure 6 illustrates the spatial–temporal variation in the EHI, IVI, and WRI for each study area. The overall annual average of each risk index in the wetlands of the main harbors of Hainan Island showed a slightly increasing trend. It shows that the ecological risk of Hainan Island is increasing in general. However, for individual harbors, none of them showed a significant trend. WKH has the lowest average EHI (0.15), followed by BAG (0.20) and DLG (0.21). CHG had the highest average EHI (0.60), followed by CQH (0.55) and BMG (0.5). The average EHI for all other ports is below 0.4.
The lowest IVI was found in CHG (0.07) and DZW (0.10). WKH (0.91) and MLG (0.34) have the highest IVI. The IVI of other harbors ranged from 0.2 to 0.3.
The average WRI was below 0.15 for all harbors, indicating that the ecological risk of coastal wetlands on Hainan Island is low. The CQH (0.15), WKH (0.14), BMG (0.12), and MLG (0.12) are of medium risk level. The average WRI for other harbors ranged from 0.05 to 0.06. DZG had the lowest mean WRI (0.03), followed by CHG (0.04) and DZW (0.05).
Figure 7 illustrates the change in the ecological risk rating of the wetlands in each study area. DZG, BAG, CHG, and DZW maintained low WRI levels throughout the study period. BMG and CQH gradually moved from low risk to high risk. WKH and DLG also moved from low risk to medium risk. In contrast, the WRI of wetlands in MLG was characterized by a more pronounced fluctuation.
Table 5 demonstrates the correlation between the main risk factors and WRI. There was a significant negative correlation between WRI and natural wetland area in Hainan Island, with a correlation coefficient of −0.703. WRI showed a significant positive correlation with the area of the aquaculture pond, with a correlation coefficient of 0.377. Wetland ecological risks on Hainan Island are more influenced by human activities. The higher the degree of protection of coastal wetlands, the lower the WRI. The stronger the human activity and the more intense the development, the higher the WRI.

4. Discussion

4.1. Uncertainty Analysis

4.1.1. Wetland Changes

In this study, the overall accuracy of the classification was greater than 90% for all study years in all study regions, which is higher than the minimum acceptable value of 85%. The kappa coefficients were all greater than 0.84, which also met the requirement of a kappa coefficient higher than 0.8. These accuracies are considered satisfactory for study areas with complex and diverse ground cover [16].
To further verify the reliability of our findings, we compared our results with those of previous studies. There was a consistent understanding of the trends in the mangrove, aquaculture pond, and tidal flats area changes. In general, areas of natural coastal wetlands, such as the mangroves on Hainan Island, did not undergo a significant decrease during the study period, while the area of artificial wetlands such as aquaculture ponds exhibited a continuous process of increase. Hu et al. [18] found that the mangrove area in China increased significantly between 1986 and 2017, but it was relatively stable in Hainan. Similar findings were reported by Wang et al. [47], Fu et al. [37], Li et al. [48], and Liao et al. [49]. Artificial land creation and aquaculture are the two most important factors causing changes in China’s coastal wetlands [50]. On Hainan Island, the rate of increase in land reclamation was low between 2000 and 2020, but the increase in aquaculture area was significant [49,50,51,52]. Spatially, aquaculture ponds are mainly located in Wenchang in the northeastern part of Hainan Island [52]. However, regionally, the changes in coastal wetlands on Hainan Island have obvious regional differences, which is also similar to the results of previous studies. Cai et al. found that no further decrease in mangrove area occurred in DZG after 1990 [23]. In contrast, Jia et al. [53] and Zhu et al. [54] showed that, between 1987 and 2020, the mangrove forests in BMG successively underwent a process of degradation followed by fluctuating recovery, but the overall mangrove area was significantly reduced, with most of the lost mangrove forests being converted into aquaculture ponds, buildings and agricultural land.
According to the results of the accuracy evaluation and the comparative analysis with previous studies, the classification accuracy of this study is relatively high. However, uncertainties and errors in impact classification results are inevitable. There are four main sources of uncertainty in this study. First, seasonal changes in vegetation phenology inevitably cause changes in vegetation spectral characteristics [16,55,56,57]. While most areas of Hainan Island are affected by clouds for many days, it is difficult to obtain images of the same phenological season. Second, some of the tidal flats or mangroves may be inundated due to the uncertainty of local transient tidal conditions at the predetermined time of satellite passage. This causes changes in the spectral characteristics of the features, resulting in the confusion of water bodies, tidal flats, and mangroves [6,55,57,58]. Third, some of the estuaries with highly turbid waters are difficult to separate from the mudflats [57]. This, together with the variation in river levels, causes uncertainty in the classification of tidal flats and water bodies. Fourth, the sensors collecting remote sensing data are different across different periods. On the one hand, it was found that the Landsat 5 TM and Landsat 7 ETM+ carry sensors with almost the same spectral band, while the Landsat 8 OLI sensor has a narrower spectral band. In areas with low vegetation cover, the NDVI calculated from Landsat 8 OLI imagery is slightly greater than that calculated from Landsat 5 TM/7 ETM+ imagery [17]. On the other hand, different sensors generate data with different spatial resolutions, thus increasing the inconsistency in the classification accuracy between different periods.

4.1.2. Risk Assessment

Although the various risk indices of the major harbor wetlands in the Hainan Island are low, the variability varies across different harbors. However, in general, the annual average WRI of Hainan Island showed a slight increasing trend. This is in agreement with previous studies. Duan et al. [36] conducted an ecological risk assessment of 35 coastal national nature reserve (NRR) wetlands across China, including four NRRs in Hainan. The results indicate that the EHI, IVI, and WRI of coastal wetlands on Hainan Island were relatively low.
This study has the advantage of using a more systematic approach to assess ecological risk changes in coastal wetlands on Hainan Island. However, the ecological risk of wetlands is influenced by many factors and with complex processes. The assessment indicators selected for this study were not comprehensive enough. Many factors that would influence the results of ecological risk assessment were not included, for example, natural factors such as sea level changes, typhoons, extreme climate events, wetland elevation changes, etc., and human factors such as the total value of national economic output related to wetland, changes in population near wetlands, changes in artificial construction area and hardened area in and around wetlands, changes in total road miles and levels in and near wetlands, artificial seawall miles, etc. If all these factors were considered, the ecological risk to wetlands on Hainan Island could be even higher; however, datasets relevant to these factors are difficult to obtain. We therefore did not take them into consideration in this study, which inevitably reduces the comprehensiveness of the assessment results [31,36,46]. The assessment system needs to be further improved to enhance the reliability of risk assessment in the future.

4.2. Trends and Potential Drivers of Wetland Change

Although extreme climatic events, changes in the quality of river inputs, biological invasions, tidal changes, and other factors can cause changes in the area and spatial distribution of coastal wetlands [29,59,60,61], sea level rise caused by global warming and increased human activities are the most important factors causing coastal wetland degradation [6,60,62]. All simulations indicate that, without further space to accommodate coastal wetland at current levels, the global coastal wetland area will decrease significantly with sea level rises [29,59,60,61].
In Hainan Island, thanks to two important policies, up to now, no significant reduction in coastal natural wetlands on Hainan Island has occurred. First, mangrove protection in Hainan Island started earlier than in other areas. Wetland reserves or wetland parks have been established in large, typical coastal wetlands on Hainan Island, such as the DZG national nature reserves, the BMG provincial nature reserve, the DZW Haiwei wetland park, etc. [23,37,53,54]. Take the DZG as an example: natural wetlands have been effectively protected there since 1986 when the DZG Mangrove National Nature Reserve was established. Therefore, the mangrove wetlands in DZG can maintain an increasing trend even though the area of aquaculture ponds continues to increase [23,37]. Although the area of aquaculture ponds has also increased significantly, this has been achieved mainly by outward expansion, and the outward expansion of artificial wetlands has, to some extent, provided the conditions for the expansion of natural wetlands. Second, Hainan Island has insisted on the strategy of regional economic development driven by tourism development for a long time. Therefore, industrial development is slow, and the damage to coastal wetlands is very low [63]. Meanwhile, the coastal wetlands on Hainan Island are facing serious threats. On the one hand, with the growth of the population and socio-economic development, a large area of coastal lowland was developed into aquaculture ponds or artificial land surface (such as hotels, tourist ports, roads, other tourist support facilities, etc.) [49,53,63]. The significant increase in elevation of the lowlands limits the space for coastal wetlands to expand inland [50,51]. On the other hand, seawall projects are very common in order to protect the personal and property safety of coastal people, which greatly limits the possibility of inland expansion of coastal wetlands [47]. Even in some of the regions with less human activity, the area available for landward migration will be much smaller than the current area of coastal wetlands, and most of this migration is at the expense of coastal freshwater wetlands [29,33]. Therefore, although there is currently no evidence of a significant decrease in the area of coastal wetlands in Hainan Island, if the intensity of human activities in coastal areas is not reduced, the area of coastal wetlands will tend to decrease in the future under the current emission scenario.

4.3. Impact of Wetland Change on Ecosystem Stability

The degradation of wetland ecosystems reduces their carbon sequestration capacity and ecological service functions [7,60,64,65]. Mangroves have a high tolerance to many of the stressors they face compared to other ecosystems [66]. However, as sea level rises, saltwater intrusion will become an increasing threat to mangroves [67]. Sea level changes will certainly affect coastal soil and hydrodynamic conditions, changing soil salinity and mangrove sedimentation [68], resulting in a forest to marsh transition, with declines in tree regeneration and growth [69]. Without sufficient setback space, the mangrove area will be reduced, habitat fragmentation will occur, and the stability of the mangrove ecosystem will be destroyed [60]. At the same time, sea level rise will change the elevation characteristics of the existing mangrove range. The elevation of the mangrove range exceeds the threshold of the elevation range that mangroves can withstand, resulting in mangrove degradation [1,30,70]. In addition, changes in soil salinity and temperature have an impact on species diversity, especially the competitiveness of native species, increasing the risk of biological invasion and reducing the stability of mangrove ecosystems [71,72,73].
Tidal wetland ecosystems are inundated by sea level rise, on the one hand, resulting in the loss of habitat for a large number of tidal organisms. On the other hand, the increasing intensity of human activities has also caused damage to the stability and ecological functions of tidal wetland ecosystems. For example, urbanization and the large-scale construction of transportation roads cause fragmentation of tidal ecosystems, while land reclamation may cut off the sea–land material exchange channels, inhibit the natural migration and deposition process of beach sediments, and accelerate the contraction and destabilization of beaches [8,20]. Land use changes under the influence of human activities, such as the conversion of wetlands and idle land to commercial and industrial land, and the conversion of wetlands to aquaculture and cropland, are the main causes of the deterioration in quality of coastal wetland habitats [62].

5. Conclusions

In this study, nine typical estuarine harbor wetlands on Hainan Island were used as the target areas for the study. Based on remote sensing data and observation data, the spatial–temporal evolution characteristics of coastal wetlands, the evolution characteristics of wetland ecological risks and the causes and ecological impacts of wetland evolution on Hainan Island from 1990 to 2020 were studied. The following conclusions are made: (1) Artificial wetlands, especially aquaculture pond areas, have increased significantly, but the total area of coastal wetlands on Hainan Island shows no significant trend. The changes in natural wetlands such as mangrove wetlands have obvious regional differences. Mangrove wetland areas with a high degree of protection, such as DZG, have increased (increased from 15.12% in 1990 to 21.89% in 2020). The mangrove wetlands in areas with strong human activities, such as BMG, were largely taken over by artificial wetlands (decreased from 15.77% in 1990 to 8.72% in 2020). (2) Overall, the coastal wetland ecosystem of Hainan Island is relatively stable and has a low ecological risk index (WRI is less than 0.15). The WRI was negatively correlated with the degree of protection and positively correlated with the intensity of human activities, with a correlation coefficient of −0.703. (3) Sea level rise and human activities are the main determinants of future trends in coastal wetlands on Hainan Island.
This study demonstrates the spatial–temporal changes and their drivers and ecological impacts in coastal wetlands on Hainan Island over the past 30 years. Sea level rise due to climate warming and increasing intensity of human activities will be the challenge for future coastal wetlands on Hainan Island to achieve their ecosystem service functions. The understanding of coastal wetland change and its drivers in Hainan Island and the key data obtained from the results of this study are a complement to the current knowledge gap in the region. These data will provide scientific guidance for regional environmental assessment and government decision-making.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/rs15041035/s1, Figure S1: Spatial distribution of coastal wetlands and AI index.

Author Contributions

Conceptualization, H.C. and D.L.; methodology, H.C.; software, H.C.; validation, H.C.; formal analysis, Y.C.; investigation, H.C., D.L. and Z.Z.; resources, H.C., D.L., Y.C. and Z.Z.; data curation, H.C. and Z.Z.; writing—original draft preparation, H.C.; writing—review and editing, H.C. and D.L.; visualization, H.C.; supervision, Y.C.; project administration, H.C. and D.L.; funding acquisition, H.C. and D.L. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Natural Science Foundation of Hainan Province, China, grant number 420QN258 and 421QN234.

Data Availability Statement

Landsat-5 TM, Landsat-7 ETM+, and Landsat-8 OLI datasets used in this study are publicly available online: https://glovis.usgs.gov/app?fullscreen=1 (accessed on 16 October 2022). Meteorological data are publicly available online: https://data.cma.cn/ (accessed on 16 October 2022). Water quality data are publicly available online: https://www.mee.gov.cn/hjzl/ (accessed on 16 October 2022) and http://hnsthb.hainan.gov.cn/ (accessed on 16 October 2022).

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Kirwan, M.L.; Megonigal, J.P. Tidal wetland stability in the face of human impacts and sea-level rise. Nature 2013, 504, 53–60. [Google Scholar] [CrossRef] [PubMed]
  2. Rodriguez, J.F.; Saco, P.M.; Sandi, S.; Saintilan, N.; Riccardi, G. Potential increase in coastal wetland vulnerability to sea-level rise suggested by considering hydrodynamic attenuation effects. Nat. Commun. 2017, 8, 16094. [Google Scholar] [CrossRef] [PubMed]
  3. Bohorquez, J.J.; Xue, G.; Frankstone, T.; Grima, M.M.; Kleinhaus, K.; Zhao, Y.; Pikitch, E.K. China’s little-known efforts to protect its marine ecosystems safeguard some habitats but omit others. Sci. Adv. 2021, 7, eabj1569. [Google Scholar] [CrossRef] [PubMed]
  4. McLeod, E.; Chmura, G.L.; Bouillon, S.; Salm, R.; Bjork, M.; Duarte, C.M.; Lovelock, C.E.; Schlesinger, W.H.; Silliman, B.R. A blueprint for blue carbon: Toward an improved understanding of the role of vegetated coastal habitats in sequestering CO2. Front. Ecol. Environ. 2011, 9, 552–560. [Google Scholar] [CrossRef] [PubMed]
  5. Wang, F.; Sanders, C.J.; Santos, I.R.; Tang, J.; Schuerch, M.; Kirwan, M.L.; Kopp, R.E.; Zhu, K.; Li, X.; Yuan, J.; et al. Global blue carbon accumulation in tidal wetlands increases with climate change. Natl. Sci. Rev. 2021, 8, nwaa296. [Google Scholar] [CrossRef]
  6. Yang, X.C.; Zhu, Z.; Qiu, S.; Kroeger, K.D.; Zhu, Z.L.; Covington, S. Detection and characterization of coastal tidal wetland change in the northeastern US using Landsat time series. Remote Sens. Environ. 2022, 276, 113171. [Google Scholar] [CrossRef]
  7. Hagger, V.; Worthington, T.A.; Lovelock, C.E.; Adame, M.F.; Amano, T.; Brown, B.M.; Friess, D.A.; Landis, E.; Mumby, P.J.; Morrison, T.H.; et al. Drivers of global mangrove loss and gain in social-ecological systems. Nat. Commun. 2022, 13, 6373. [Google Scholar] [CrossRef]
  8. Murray, N.J.; Phinn, S.R.; DeWitt, M.; Ferrari, R.; Johnston, R.; Lyons, M.B.; Clinton, N.; Thau, D.; Fuller, R.A. The global distribution and trajectory of tidal flats. Nature 2019, 565, 222–225. [Google Scholar] [CrossRef]
  9. Richards, D.R.; Thompson, B.S.; Wijedasa, L. Quantifying net loss of global mangrove carbon stocks from 20 years of land cover change. Nat. Commun. 2020, 11, 4260. [Google Scholar] [CrossRef]
  10. Frederikse, T.; Landerer, F.; Caron, L.; Adhikari, S.; Parkes, D.; Humphrey, V.W.; Dangendorf, S.; Hogarth, P.; Zanna, L.; Cheng, L.; et al. The causes of sea-level rise since 1900. Nature 2020, 584, 393–397. [Google Scholar] [CrossRef] [PubMed]
  11. Walker, J.S.; Kopp, R.E.; Little, C.M.; Horton, B.P. Timing of emergence of modern rates of sea-level rise by 1863. Nat. Commun. 2022, 13, 966. [Google Scholar] [CrossRef]
  12. Murray, N.J.; Worthington, T.A.; Bunting, P.; Duce, S.; Hagger, V.; Lovelock, C.E.; Lucas, R.; Saunders, M.I.; Sheaves, M.; Spalding, M.; et al. High-resolution mapping of losses and gains of Earth’s tidal wetlands. Science 2022, 376, 744–749. [Google Scholar] [CrossRef]
  13. Guo, M.; Li, J.; Sheng, C.; Xu, J.; Wu, L. A Review of Wetland Remote Sensing. Remote Sens. 2017, 17, 777. [Google Scholar] [CrossRef] [PubMed]
  14. Pham, T.D.; Yokoya, N.; Bui, D.T.; Yoshino, K.; Friess, D.A. Remote Sensing Approaches for Monitoring Mangrove Species, Structure, and Biomass: Opportunities and Challenges. Remote Sens. 2019, 11, 230. [Google Scholar] [CrossRef]
  15. Wang, J.; Chen, J.S.; Wen, Y.; Fan, W.; Liu, Q.N.; Tarolli, P. Monitoring the coastal wetlands dynamics in Northeast Italy from 1984 to 2016. Ecol. Indic. 2021, 129, 107906. [Google Scholar] [CrossRef]
  16. Dang, A.T.N.; Kumar, L.; Reid, M.; Nguyen, H. Remote Sensing Approach for Monitoring Coastal Wetland in the Mekong Delta, Vietnam: Change Trends and Their Driving Forces. Remote Sens. 2021, 13, 3359. [Google Scholar] [CrossRef]
  17. He, K.; Zhang, Y.; Li, W.; Sun, G.; McNulty, S. Detecting Coastal Wetland Degradation by Combining Remote Sensing and Hydrologic Modeling. Forests 2022, 13, 411. [Google Scholar] [CrossRef]
  18. Hu, L.; Li, W.; Xu, B. Monitoring mangrove forest change in China from 1990 to 2015 using Landsat-derived spectral-temporal variability metrics. Int. J. Appl. Earth Obs. Geoinf. 2018, 73, 88–98. [Google Scholar] [CrossRef]
  19. Jia, M.; Wang, Z.; Zhang, Y.; Mao, D.; Wang, C. Monitoring loss and recovery of mangrove forests during 42 years: The achievements of mangrove conservation in China. Int. J. Appl. Earth Obs. Geoinf. 2018, 73, 535–545. [Google Scholar] [CrossRef]
  20. Li, Y.; Mao, D.; Wang, Z.; Wang, X.; Tan, X.; Jia, M.; Ren, C. Identifying variable changes in wetlands and their anthropogenic threats bordering the Yellow Sea for water bird conservation. Glob. Ecol. Conserv. 2021, 27, e01613. [Google Scholar] [CrossRef]
  21. Chen, C.; Feng, J.; Wang, C.; Mao, L.; Zhang, Y. Satellite-Based Monitoring of Coastal Wetlands in Yancheng, Jiangsu Province, China. J. Mar. Sci. Eng. 2022, 10, 829. [Google Scholar] [CrossRef]
  22. Yan, J.; Du, J.; Su, F.; Zhao, S.; Zhang, S.; Feng, P. Reclamation and Ecological Service Value Evaluation of Coastal Wetlands Using Multispectral Satellite Imagery. Wetlands 2022, 42, 1–15. [Google Scholar] [CrossRef]
  23. Cai, R.; Ding, R.; Yan, X.; Li, C.; Sun, J.; Tan, H.; Men, W.; Guo, H.; Wang, C. Adaptive response of Dongzhaigang mangrove in China to future sea level rise. Sci. Rep. 2022, 12, 11495. [Google Scholar] [CrossRef]
  24. Aslan, A.; Rahman, A.F.; Warren, M.W.; Robeson, S.M. Mapping spatial distribution and biomass of coastal wetland vegetation in Indonesian Papua by combining active and passive remotely sensed data. Remote Sens. Environ. 2016, 183, 65–81. [Google Scholar] [CrossRef]
  25. Zhang, A.; Sun, G.; Ma, P.; Jia, X.; Ren, J.; Huang, H.; Zhang, X. Coastal Wetland Mapping with Sentinel-2 MSI Imagery Based on Gravitational Optimized Multilayer Perceptron and Morphological Attribute Profiles. Remote Sens. 2019, 11, 952. [Google Scholar] [CrossRef]
  26. Chang, M.H.; Meng, X.C.; Sun, W.W.; Yang, G.; Peng, J.T. Collaborative Coupled Hyperspectral Unmixing Based Subpixel Change Detection for Analyzing Coastal Wetlands. Ieee J. Sel. Top. Appl. Earth Obs. Remote Sens. 2021, 14, 8208–8224. [Google Scholar] [CrossRef]
  27. Liu, J.; Li, P.; Tu, C.; Wang, H.; Zhou, Z.; Feng, Z.; Shen, F.; Li, Z. Spatiotemporal Change Detection of Coastal Wetlands Using Multi-Band SAR Coherence and Synergetic Classification. Remote Sens. 2022, 14, 2610. [Google Scholar] [CrossRef]
  28. Yuan, Y.; Meng, X.; Sun, W.; Yang, G.; Wang, L.; Peng, J.; Wang, Y. Multi-Resolution Collaborative Fusion of SAR, Multispectral and Hyperspectral Images for Coastal Wetlands Mapping. Remote Sens. 2022, 14, 3492. [Google Scholar] [CrossRef]
  29. Schuerch, M.; Spencer, T.; Temmerman, S.; Kirwan, M.L.; Wolff, C.; Lincke, D.; McOwen, C.J.; Pickering, M.D.; Reef, R.; Vafeidis, A.T.; et al. Future response of global coastal wetlands to sea-level rise. Nature 2018, 561, 231–234. [Google Scholar] [CrossRef]
  30. Saintilan, N.; Khan, N.S.; Ashe, E.; Kelleway, J.J.; Rogers, K.; Woodroffe, C.D.; Horton, B.P. Thresholds of mangrove survival under rapid sea level rise. Science 2020, 368, 1118–1121. [Google Scholar] [CrossRef] [PubMed]
  31. Estrela-Segrelles, C.; Gomez-Martinez, G.; Perez-Martin, M.A. Risk assessment of climate change impacts on Mediterranean coastal wetlands. Application in Jucar River Basin District (Spain). Sci. Total Environ. 2021, 790, 13. [Google Scholar] [CrossRef] [PubMed]
  32. Chatting, M.; Al-Maslamani, I.; Walton, M.; Skov, M.W.; Kennedy, H.; Husrevoglu, Y.S.; Le Vay, L. Future Mangrove Carbon Storage Under Climate Change and Deforestation. Front. Mar. Sci. 2022, 9, 58. [Google Scholar] [CrossRef]
  33. Osland, M.J.; Chivoiu, B.; Enwright, N.M.; Thorne, K.M.; Guntenspergen, G.R.; Grace, J.B.; Dale, L.L.; Brooks, W.; Herold, N.; Day, J.W.; et al. Migration and transformation of coastal wetlands in response to rising seas. Sci. Adv. 2022, 8, eabo5174. [Google Scholar] [CrossRef] [PubMed]
  34. Van der Stocken, T.; Vanschoenwinkel, B.; Carroll, D.; Cavanaugh, K.C.; Koedam, N. Mangrove dispersal disrupted by projected changes in global seawater density. Nat. Clim. Change 2022, 12, 685–691. [Google Scholar] [CrossRef]
  35. Li, H.; Li, L.; Su, F.; Wang, T.; Gao, P. Ecological stability evaluation of tidal flat in coastal estuary: A case study of Liaohe estuary wetland, China. Ecol. Indic. 2021, 130, 108032. [Google Scholar] [CrossRef]
  36. Duan, H.; Yu, X.; Zhang, L.; Xia, S.; Liu, Y.; Mao, D.; Zhang, G. An evaluating system for wetland ecological risk: Case study in coastal mainland China. Sci. Total Environ. 2022, 828, 154535. [Google Scholar] [CrossRef]
  37. Fu, C.; Song, X.Q.; Xie, Y.; Wang, C.; Luo, J.B.; Fang, Y.; Cao, B.; Qiu, Z.X. Research on the Spatiotemporal Evolution of Mangrove Forests in the Hainan Island from 1991 to 2021 Based on SVM and Res-UNet Algorithms. Remote Sens. 2022, 14, 5554. [Google Scholar] [CrossRef]
  38. Candiago, S.; Remondino, F.; De Giglio, M.; Dubbini, M.; Gattelli, M. Evaluating Multispectral Images and Vegetation Indices for Precision Farming Applications from UAV Images. Remote Sens. 2015, 7, 4026–4047. [Google Scholar] [CrossRef]
  39. Gao, B.C. NDWI—A normalized difference water index for remote sensing of vegetation liquid water from space. Remote Sens. Environ. 1996, 58, 257–266. [Google Scholar] [CrossRef]
  40. Xu, H.Q. Modification of normalised difference water index (NDWI) to enhance open water features in remotely sensed imagery. Int. J. Remote Sens. 2006, 27, 3025–3033. [Google Scholar] [CrossRef]
  41. Gupta, K.; Mukhopadhyay, A.; Giri, S.; Chanda, A.; Majumdar, S.D.; Samanta, S.; Mitra, D.; Samal, R.N.; Pattnaik, A.K.; Hazra, S. An index for discrimination of mangroves from non-mangroves using LANDSAT 8 OLI imagery. Methodsx 2018, 5, 1129–1139. [Google Scholar] [CrossRef] [PubMed]
  42. Matthews, G.V.T. The Ramsar Convention on Wetlands: Its History and Development; Ramsar Convention Bureau: Gland, Switzerland, 1993; pp. i–vi, 1–120. [Google Scholar]
  43. Dragut, L.; Tiede, D.; Levick, S.R. ESP: A tool to estimate scale parameter for multiresolution image segmentation of remotely sensed data. Int. J. Geogr. Inf. Sci. 2010, 24, 859–871. [Google Scholar] [CrossRef]
  44. Dragut, L.; Csillik, O.; Eisank, C.; Tiede, D. Automated parameterisation for multi-scale image segmentation on multiple layers. ISPRS-J. Photogramm. Remote Sens. 2014, 88, 119–127. [Google Scholar] [CrossRef] [PubMed]
  45. Foody, G.M. Status of land cover classification accuracy assessment. Remote Sens. Environ. 2002, 80, 185–201. [Google Scholar] [CrossRef]
  46. Li, Z.; Jiang, W.; Wang, W.; Chen, Z.; Ling, Z.; Lv, J. Ecological risk assessment of the wetlands in Beijing-Tianjin-Hebei urban agglomeration. Ecol. Indic. 2020, 117, 106677. [Google Scholar] [CrossRef]
  47. Wang, X.; Xiao, X.; Xu, X.; Zou, Z.; Chen, B.; Qin, Y.; Zhang, X.; Dong, J.; Liu, D.; Pan, L.; et al. Rebound in China’s coastal wetlands following conservation and restoration. Nat. Sustain. 2021, 4, 1076–1083. [Google Scholar] [CrossRef]
  48. Li, Y.; Wen, H.; Wang, F. Analysis of the Evolution of Mangrove Landscape Patterns and Their Drivers in Hainan Island from 2000 to 2020. Sustainability 2023, 15, 759. [Google Scholar] [CrossRef]
  49. Liao, J.J.; Zhen, J.N.; Zhang, L.; Metternicht, G. Understanding Dynamics of Mangrove Forest on Protected Areas of Hainan Island, China: 30 Years of Evidence from Remote Sensing. Sustainability 2019, 11, 5356. [Google Scholar] [CrossRef]
  50. Jiang, S.; Xu, N.; Li, Z.C.; Huang, C.H. Satellite derived coastal reclamation expansion in China since the 21st century. Glob. Ecol. Conserv. 2021, 30, e01797. [Google Scholar] [CrossRef]
  51. Herbeck, L.S.; Krumme, U.; Andersen, T.J.; Jennerjahn, T.C. Decadal trends in mangrove and pond aquaculture cover on Hainan (China) since 1966: Mangrove loss, fragmentation and associated biogeochemical changes. Estuar. Coast. Shelf Sci. 2020, 233, 106531. [Google Scholar] [CrossRef]
  52. Fu, T.M.; Zhang, L.; Yuan, X.; Chen, B.W.; Yan, M. Spatio-temporal patterns and sustainable development of coastal aquaculture in Hainan Island, China: 30 Years of evidence from remote sensing. Ocean. Coast. Manag. 2021, 214, 12. [Google Scholar] [CrossRef]
  53. Jia, P.H.; Huang, W.D.; Zhang, Z.Y.; Cheng, J.X.; Xiao, Y.L. The Carbon Sink of Mangrove Ecological Restoration between 1988–2020 in Qinglan Bay, Hainan Island, China. Forests 2022, 13, 1547. [Google Scholar] [CrossRef]
  54. Zhu, B.; Liao, J.J.; Shen, G.Z. Combining time series and land cover data for analyzing spatio-temporal changes in mangrove forests: A case study of Qinglangang Nature Reserve, Hainan, China. Ecol. Indic. 2021, 131, 16. [Google Scholar] [CrossRef]
  55. Pham, H.N.; Dang, K.B.; Nguyen, T.V.; Tran, N.C.; Ngo, X.Q.; Nguyen, D.A.; Phan, T.T.H.; Nguyen, T.T.; Guo, W.; Ngo, H.H. A new deep learning approach based on bilateral semantic segmentation models for sustainable estuarine wetland ecosystem management. Sci. Total Environ. 2022, 838, 155826. [Google Scholar] [CrossRef] [PubMed]
  56. Jia, M.; Mao, D.; Wang, Z.; Ren, C.; Zhu, Q.; Li, X.; Zhang, Y. Tracking long-term floodplain wetland changes: A case study in the China side of the Amur River Basin. Int. J. Appl. Earth Obs. Geoinf. 2020, 92, 102185. [Google Scholar] [CrossRef]
  57. Jia, M.; Wang, Z.; Mao, D.; Ren, C.; Wang, C.; Wang, Y. Rapid, robust, and automated mapping of tidal flats in China using time series Sentinel-2 images and Google Earth Engine. Remote Sens. Environ. 2021, 255, 112285. [Google Scholar] [CrossRef]
  58. Tian, Y.; Jia, M.; Wang, Z.; Mao, D.; Du, B.; Wang, C. Monitoring Invasion Process of Spartina alterniflora by Seasonal Sentinel-2 Imagery and an Object-Based Random Forest Classification. Remote Sens. 2020, 12, 1383. [Google Scholar] [CrossRef]
  59. DeConto, R.M.; Pollard, D.; Alley, R.B.; Velicogna, I.; Gasson, E.; Gomez, N.; Sadai, S.; Condron, A.; Gilford, D.M.; Ashe, E.L.; et al. The Paris Climate Agreement and future sea-level rise from Antarctica. Nature 2021, 593, 83–89. [Google Scholar] [CrossRef]
  60. Fant, C.; Gentile, L.E.; Herold, N.; Kunkle, H.; Kerrich, Z.; Neumann, J.; Martinich, J. Valuation of long-term coastal wetland changes in the U.S. Ocean. Coast. Manag. 2022, 226, 106248. [Google Scholar] [CrossRef]
  61. Visschers, L.L.B.; Santos, C.D.; Franco, A.M.A. Accelerated migration of mangroves indicate large-scale saltwater intrusion in Amazon coastal wetlands. Sci. Total Environ. 2022, 836, 8. [Google Scholar] [CrossRef] [PubMed]
  62. Zhang, X.; Song, W.; Lang, Y.; Feng, X.; Yuan, Q.; Wang, J. Land use changes in the coastal zone of China’s Hebei Province and the corresponding impacts on habitat quality. Land Use Policy 2020, 99, 104957. [Google Scholar] [CrossRef]
  63. Chu, L.X.; Oloo, F.; Chen, B.; Xie, M.M.; Blaschke, T. Assessing the Influence of Tourism-Driven Activities on Environmental Variables on Hainan Island, China. Remote Sens. 2020, 12, 2813. [Google Scholar] [CrossRef]
  64. Hagger, V.; Waltham, N.J.; Lovelock, C.E. Opportunities for coastal wetland restoration for blue carbon with co-benefits for biodiversity, coastal fisheries, and water quality. Ecosyst. Serv. 2022, 55, 101423. [Google Scholar] [CrossRef]
  65. Iram, N.; Maher, D.T.; Lovelock, C.E.; Baker, T.; Cadier, C.; Adame, M.F. Climate change mitigation and improvement of water quality from the restoration of a subtropical coastal wetland. Ecol. Appl. 2022, 32, e2620. [Google Scholar] [CrossRef] [PubMed]
  66. Aljahdali, M.O.; Munawar, S.; Khan, W.R. Monitoring Mangrove Forest Degradation and Regeneration: Landsat Time Series Analysis of Moisture and Vegetation Indices at Rabigh Lagoon, Red Sea. Forests 2021, 12, 52. [Google Scholar] [CrossRef]
  67. Gilani, H.; Naz, H.I.; Arshad, M.; Nazim, K.; Akram, U.; Abrar, A.; Asif, M. Evaluating mangrove conservation and sustainability through spatiotemporal (1990-2020) mangrove cover change analysis in Pakistan. Estuar. Coast. Shelf Sci. 2021, 249, 16. [Google Scholar] [CrossRef]
  68. Hanggara, B.B.; Murdiyarso, D.; Ginting, Y.R.S.; Widha, Y.L.; Panjaitan, G.Y.; Lubis, A.A. Effects of diverse mangrove management practices on forest structure, carbon dynamics and sedimentation in North Sumatra, Indonesia. Estuar. Coast. Shelf Sci. 2021, 259, 107467. [Google Scholar] [CrossRef]
  69. Conner, W.; Whitmire, S.; Duberstein, J.; Stalter, R.; Baden, J. Changes within a South Carolina Coastal Wetland Forest in the Face of Rising Sea Level. Forests 2022, 13, 414. [Google Scholar] [CrossRef]
  70. Xiong, Y.; Jiang, Z.; Xin, K.; Liao, B.; Chen, Y.; Li, M.; Guo, H.; Xu, Y.; Zhai, X.; Zhang, C. Factors influencing mangrove forest recruitment in rehabilitated aquaculture ponds. Ecol. Eng. 2021, 168, 106272. [Google Scholar] [CrossRef]
  71. BORDE, A.B.; DIEFENDERFER, H.L.; CULLINAN, V.I.; ZIMMERMAN, S.A.; THOM, R.M. Ecohydrology of wetland plant communities along an estuarine to tidal river gradient. Ecoshpere 2020, 11, e03185. [Google Scholar] [CrossRef]
  72. Ren, J.; Chen, J.; Xu, C.; van de Koppel, J.; Thomsen, M.S.; Qiu, S.; Cheng, F.; Song, W.; Liu, Q.-X.; Xu, C.; et al. An invasive species erodes the performance of coastal wetland protected areas. Sci. Adv. 2021, 7, eabi8943. [Google Scholar] [CrossRef] [PubMed]
  73. Martínez-Megías, C.; Rico, A. Biodiversity impacts by multiple anthropogenic stressors in Mediterranean coastal wetlands. Sci. Total Environ. 2022, 818, 151712. [Google Scholar] [CrossRef] [PubMed]
Remotesensing 15 01035 g001 550
Figure 1. Map of the study area.
Figure 1. Map of the study area.
Remotesensing 15 01035 g001
Remotesensing 15 01035 g002 550
Figure 2. Comparison of object-oriented segmentation. (a) ESP segmentation; (b) segmentation scale of 100; (c) segmentation scale of 50.
Figure 2. Comparison of object-oriented segmentation. (a) ESP segmentation; (b) segmentation scale of 100; (c) segmentation scale of 50.
Remotesensing 15 01035 g002
Remotesensing 15 01035 g003 550
Figure 3. Comparison of Google Earth map and Landsat image: ((a-1)–(c-1)) are the 2020 Google Earth image of CQH, DZW, and BAG, respectively; ((a-2)–(c-2)) are the 2005 Landsat TM image of CQH, DZW, and BAG, respectively; ((a-3)–(c-3)) are the 2020 Landsat OLI image the of CQH, DZW, and BAG, respectively.
Figure 3. Comparison of Google Earth map and Landsat image: ((a-1)–(c-1)) are the 2020 Google Earth image of CQH, DZW, and BAG, respectively; ((a-2)–(c-2)) are the 2005 Landsat TM image of CQH, DZW, and BAG, respectively; ((a-3)–(c-3)) are the 2020 Landsat OLI image the of CQH, DZW, and BAG, respectively.
Remotesensing 15 01035 g003
Remotesensing 15 01035 g004 550
Figure 4. (a) Change in area of mangrove wetlands; (b) change in area of artificial wetlands (aquaculture ponds); (c) change in area of natural wetlands; (d) change in area of all wetlands.
Figure 4. (a) Change in area of mangrove wetlands; (b) change in area of artificial wetlands (aquaculture ponds); (c) change in area of natural wetlands; (d) change in area of all wetlands.
Remotesensing 15 01035 g004
Remotesensing 15 01035 g005 550
Figure 5. Spatial–temporal variation of the ratios of (a) mangrove, (b) aquaculture pond, (c) natural wetland and (d) all wetland to the total areas of the study areas.
Figure 5. Spatial–temporal variation of the ratios of (a) mangrove, (b) aquaculture pond, (c) natural wetland and (d) all wetland to the total areas of the study areas.
Remotesensing 15 01035 g005
Remotesensing 15 01035 g006 550
Figure 6. (a) External hazard index (EHI); (b) internal vulnerability index (IVI); (c) wetland risk index (WRI).
Figure 6. (a) External hazard index (EHI); (b) internal vulnerability index (IVI); (c) wetland risk index (WRI).
Remotesensing 15 01035 g006
Remotesensing 15 01035 g007 550
Figure 7. Wetland risk classification.
Figure 7. Wetland risk classification.
Remotesensing 15 01035 g007
Table
Table 1. Indices used for land use classification.
Table 1. Indices used for land use classification.
IndexCalculation FormulaDescription
NDVINDVI = (Nir − Red)/(Nir + Red)Detection of vegetation growth status, vegetation cover, and partial elimination of radiation errors.
NDWINDWI = (Green − Nir)/(Green + reed)Used to extract water body information from images with good effect.
MNDWIMNDWI = (Green − Swir1)/(Green + Swir2)MNDWI is better than NDWI for revealing the microfeatures of water bodies and it can easily distinguish shadows from water bodies, solving the difficulty of eliminating shadows in water body extraction.
CMRICMRI = NDVI − NDWICMRI improves the identification of mangrove and non-mangrove vegetation.
Table
Table 2. Classification system.
Table 2. Classification system.
Category ICategory IICodeDescriptionLandsat Images
Natural WetlandWater11Natural water bodies, including rivers, lakes, and oceans.Remotesensing 15 01035 i001
Mangrove12Mangrove forest.Remotesensing 15 01035 i002
Tidal flat13131Sandy beaches (sandy beaches also belong to tidal flats, but the spectral characteristics of sandy beaches are not consistent with other tidal flats, so they are listed separately).Remotesensing 15 01035 i003
132Tidal flats, mainly bare wetlands in natural wetlands other than water bodies, mangroves, and sandy beaches.Remotesensing 15 01035 i004
Artificial wetlandAquaculture pond21Aquaculture pond.Remotesensing 15 01035 i005
Non-wetland landscapeTerrestrial vegetation31Terrestrial vegetation.Remotesensing 15 01035 i006
Other32Other terrestrial landscapes.Remotesensing 15 01035 i007
Table
Table 3. Landscape distribution indices.
Table 3. Landscape distribution indices.
IndexCalculation FormulaDescription
Landscape fragmentation index (C) C = N i / A i N i denotes the total number of patches in class i landscapes. A i denotes the total area of class i landscape. The larger C is, the higher the fragmentation of landscape distribution.
Aggregation index (AI) A I = g i i / m a x → g i i g i i refers to the number of adjacent patches in the landscape. The higher the value of AI, the higher the degree of aggregation. A larger value of AI means that the landscape consists of a few clustered large patches. A small value of AI means that the landscape consists of many small patches.
Table
Table 4. Classification accuracy assessment.
Table 4. Classification accuracy assessment.
Site 1990199520002005201020152020
DZGOverall accuracy (%)91.190.790.991.892.293.494.3
Kappa coefficient0.850.840.860.890.900.920.91
BMGOverall accuracy (%)90.791.392.791.591.992.593.7
Kappa coefficient0.860.850.870.880.890.910.90
BAGOverall accuracy (%)90.091.892.990.190.293.594.4
Kappa coefficient0.890.870.880.900.920.940.95
CHGOverall accuracy (%)90.890.591.192.692.592.193.2
Kappa coefficient0.870.890.920.880.900.930.92
DZWOverall accuracy (%)91.292.590.790.991.194.295.3
Kappa coefficient0.880.860.840.890.900.910.93
WKHOverall accuracy (%)90.291.590.391.191.493.594.1
Kappa coefficient0.850.870.860.890.880.910.92
DLGOverall accuracy (%)90.190.591.391.792.291.393.4
Kappa coefficient0.870.880.860.850.880.920.91
MLGOverall accuracy (%)92.292.191.992.891.692.194.0
Kappa coefficient0.880.850.870.860.890.910.93
CQHOverall accuracy (%)90.890.691.190.891.294.695.4
Kappa coefficient0.880.850.870.860.900.920.94
Table
Table 5. Correlation between the main risk factors and WRI.
Table 5. Correlation between the main risk factors and WRI.
PrecipitationTemperatureWater QualityAquiculture PondNature WetlandWRI
Precipitation1.0000.0010.1060.113−0.143−0.067
Temperature 1.0000.429 **0.291 *0.0120.086
Water quality 1.0000.411 **0.0100.200
Aquiculture pond 1.000−0.1590.377 **
Nature wetland 1.000−0.703 **
WRI 1.000
Note: ** Significant correlation at the 0.01 level (two-tailed). * Correlation is significant at the 0.05 level (two-tailed).
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Chen, H.; Li, D.; Chen, Y.; Zhao, Z. Spatial–Temporal Evolution Monitoring and Ecological Risk Assessment of Coastal Wetlands on Hainan Island, China. Remote Sens. 2023, 15, 1035. https://doi.org/10.3390/rs15041035

AMA Style

Chen H, Li D, Chen Y, Zhao Z. Spatial–Temporal Evolution Monitoring and Ecological Risk Assessment of Coastal Wetlands on Hainan Island, China. Remote Sensing. 2023; 15(4):1035. https://doi.org/10.3390/rs15041035

Chicago/Turabian Style

Chen, Haiyan, Dalong Li, Yaning Chen, and Zhizhong Zhao. 2023. "Spatial–Temporal Evolution Monitoring and Ecological Risk Assessment of Coastal Wetlands on Hainan Island, China" Remote Sensing 15, no. 4: 1035. https://doi.org/10.3390/rs15041035

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop