Figure - available from: Remote Sensing in Ecology and Conservation
This content is subject to copyright. Terms and conditions apply.
Example coral reef habitat maps on the Great Barrier Reef showing differences in spatial resolution from different satellite sensors for a geomorphic map of Heron Reef (panels A–D; red square in Figure 2 panel C) and a benthic map of Batt Reef (panels E–H; red square in Figure 2 panel B). The maps can be explored in detail here: mitchest.users.earthengine.app/view/coral-map-explorer.
Source publication
Coral reefs are among the most diverse and iconic ecosystems on Earth, but a range of anthropogenic pressures are threatening their persistence. Owing to their remoteness, broad spatial coverage and cross‐jurisdictional locations, there are no high‐resolution remotely sensed maps available at the global scale. Here we present a framework that is ca...
Similar publications
Coral reef is one of the coastal ecosystems that has an important role as a place to look for food, shelter, and breeding grounds for other biotas. Besides, the ecological aspects of coral reefs as a coastal protectors from waves and abrasion. CPCe (Coral Points Count with Excel extension) is used to observe or monitoring reefs ecpsystems. The phot...
Citations
... In particular, high-resolution remote sensing also has high positioning accuracy. Therefore, remote sensing technology is proven to be an effective technical means for coral islands and reefs monitoring [8][9][10][11][12], and remote sensing data with multiple platforms, sensors, and spatial and spectral resolutions are applied to coral islands and reefs research [13][14][15][16][17][18]. Remote sensing research on coral islands and reefs focuses on analyzing the spatiotemporal dynamic changes of coral islands [19][20][21][22][23][24][25][26][27], evaluating the stability of coral sandbanks [28][29][30], and mapping the geomorphology and habitats of coral reefs [31][32][33][34][35][36][37][38]. ...
Coral islands and reefs are formed by the cementation of the remains of shallow water reef-building coral polyps and other reef dwelling organisms in tropical oceans. They can be divided into coral islands, coral sandbanks, coral reefs, and coral shoals, of which, Coral shoals are located below the depth datum and are not exposed even at low tide, and sometimes are distributed at water depths exceeding 30 m. Satellite images with wide spatial–temporal coverage have played a crucial role in coral island and reef monitoring, and remote sensing data with multiple platforms, sensors, and spatial and spectral resolutions are employed. However, the accurate detection of coral shoals remains challenging mainly due to the depth effect, that is, coral shoals, especially deeper ones, have very similar spectral characteristics to the sea in optical images. Here, an optical remote sensing detection method is proposed to rapidly and accurately detect the coral shoals using a deep belief network (DBN) from optical satellite imagery. The median filter is used to filter the DBN classification results, and the appropriate filtering window is selected according to the spatial resolution of the optical images. The proposed method demonstrated outstanding performance by validating and comparing the detection results of the Yinli Shoal. Moreover, the expected results are obtained by applying this method to other coral shoals in the Xisha Islands, including the Binmei Shoal, Beibianlang, Zhanhan Shoal, Shanhudong Shoal, and Yongnan Shoal. This detection method is expected to provide the coral shoals’ information rapidly once optical satellite images are available and cloud cover and tropical cyclones are satisfactory. The further integration of the detection results of coral shoals with water depth and other information can effectively ensure the safe navigation of ships.
... Deep learning methods can handle better the complex interaction of light with the water surface, column, and water bottom compared to simple models [3]. On the other hand, deep learning-based pixel classification approaches of the seabed are mainly devoted to map coral reefs or seagrass meadows [10,11] and mainly involve fully convolutional network (FCN) architectures. The existing methods for both SDB and pixel classification are mostly applied on non-public dataset. ...
Accurate, detailed, and high-frequent bathymetry, coupled with complex semantic content, is crucial for the undermapped shallow seabed areas facing intense climatological and anthropogenic pressures. Current methods exploiting remote sensing images to derive bathymetry or seabed classes mainly exploit non-open data. This lack of openly accessible benchmark archives prevents the wider use of deep learning methods in such applications. To address this issue, in this paper we present the MagicBathyNet, which is a benchmark dataset made up of image patches of Sentinel2, SPOT-6 and aerial imagery, bathymetry in raster format and annotations of seabed classes. MagicBathyNet is then exploited to benchmark state-of-the-art methods in learning-based bathymetry and pixel-based classification. Dataset, pre-trained weights, and code are publicly available at www.magicbathy.eu/magicbathynet.html.
... For example, monthly composites, rather than annual composites used here, could be produced to investigate intra-annual changes in seagrass extent. We expect our workflow to be widely applicable over large spatial extents, as the structure of our method is modelled on previous work that has mapped contemporary seagrass area across 113,000 km 216 and, by implementing regional segmentation, across the full extent of the tropics 18 . The training and validation of such models are crucially important steps which should involve experts with first-hand knowledge and experience of the region mapped. ...
... To produce a national contemporary seagrass map for the Maldives, a workflow was adapted using established methods 17,18 . The satellite data used to produce the contemporary maps consisted of Sentinel-2 images which were retrieved and pre-processed in GEE following the standard protocol for aquatic remote sensing of benthic habitats (see full details in supplementary material). ...
... Representative training polygons were generated across 3 classes, seagrass, non-seagrass (which included coral reefs, mangroves, sand/ rubble, and macroalgal beds), and optical-deep water (ODW), to sample pixel-level data on the spectral, elevation, and slope values of each class. The training data were labelled over the full extent of the Maldives, and across the full range of geomorphological features and water depths, as informed by the bathymetric DEM (10.5 × 10.5 m pixel size) and Allen Coral Atlas geomorphic thematic map (5 × 5 m pixel size) 18 . In total 25,463 training pixels were generated (n = 5761 seagrass; n = 15,433 non-seagrass; n = 4269 ODW; Fig. 1). ...
The areal extent of seagrass meadows is in rapid global decline, yet they provide highly valuable societal benefits. However, their conservation is hindered by data gaps on current and historic spatial extents. Here, we outline an approach for national-scale seagrass mapping and monitoring using an open-source platform (Google Earth Engine) and freely available satellite data (Landsat, Sentinel-2) that can be readily applied in other countries globally. Specifically, we map contemporary (2021) and historical (2000–2021; n = 10 maps) shallow water seagrass extent across the Maldives. We found contemporary Maldivian seagrass extent was ~ 105 km² (overall accuracy = 82.04%) and, notably, that seagrass area increased threefold between 2000 and 2021 (linear model, + 4.6 km² year⁻¹, r² = 0.93, p < 0.001). There was a strongly significant association between seagrass and anthropogenic activity (p < 0.001) that we hypothesize to be driven by nutrient loading and/or altered sediment dynamics (from large scale land reclamation), which would represent a beneficial anthropogenic influence on Maldivian seagrass meadows. National-scale tropical seagrass expansion is unique against the backdrop of global seagrass decline and we therefore highlight the Maldives as a rare global seagrass ‘bright spot’ highly worthy of increased attention across scientific, commercial, and conservation policy contexts.
... These classes are used globally for all reef. When using 2-5-m image resolution, high accuracy values (0.8-0.9 overall accuracies) were reported for Australian reefs and for 'Pacific Ocean reefs' by Lyons et al. (2020), although validation data seem to come also from other maps and not just from actual field data (Lyons et al. 2020;Roelfsema et al. 2021). In this case, using ancillary maps cannot be really considered as actual validation data. ...
... These classes are used globally for all reef. When using 2-5-m image resolution, high accuracy values (0.8-0.9 overall accuracies) were reported for Australian reefs and for 'Pacific Ocean reefs' by Lyons et al. (2020), although validation data seem to come also from other maps and not just from actual field data (Lyons et al. 2020;Roelfsema et al. 2021). In this case, using ancillary maps cannot be really considered as actual validation data. ...
... This error was easily detected, as seagrass species that can accommodate high energy environment like Thalassadendron ciliatum are absent from New Caledonia . It is surprising that such errors occur as they seem to be handled by a specific contextual rule (Lyons et al. 2020). The small sizes of these patches make them minor errors though. ...
Several remote sensing projects have produced coral reef habitat maps globally. This includes the Allen Coral Atlas (ACA) project. The present study focuses on assessing the accuracy of the ACA 6-class benthic habitat products available for New Caledonia in the 0–10-m depth range. The assessment involves using independent control data collected through thousands of in situ photographs of the benthic environment during sea cucumber stock assessments on multiple sites around New Caledonia. The overall accuracy of the ACA benthic product ranked from 20 to 67% depending on the sites and evaluation method. These results are generally lower than those achieved by other global and local coral reef mapping and products which can be primarily explained by the lack of local ground-truth data to train the classifications. In addition, we discuss other significant misclassifications immediately detectable through historical knowledge, and the limited relevance of the classification scheme to represent the actual diversity of New Caledonia coral reef habitats, a key criteria to guide conservation with habitat maps. Overall, the study highlights the limitations of the ACA benthic product that users should be aware of and offer some recommendations and caveats for both potential users and map producers in other areas.
... e field of MPP and associated severe impact of MPs on the coral ecosystem (Fig. 3A). Corals and coral reefs are found more or less in different oceanic beds of tropical and subtropical regions, but these are intensively clustered in the Southeast Asian Ocean, Pacific, Atlantic Ocean, and Indian Oceans surrounded by Australia (Andrello et al., 2022;B. Lyons et al., 2020). Researchers in different countries worked with coral, and MPs impacted the coral reef ecosystem (Fig. 3B). ...
Microplastics (MPs) are one of the leading pollutants on the earth's surface and are found almost everywhere, including in aquatic environments. Microplastic pollution (MPP) on the oceanic surface is causing widespread concern among scientists and researchers worldwide due to its life-threatening impact on underwater living organisms. Excessive production of MP components and unscientific disposal caused severe issues for the marine ecosystem. Invertebrate corals and coral reefs in tropical and subtropical countries have suffered immensely from MP pollution and have shown a gradually decreasing trend of coral reef concentration on the oceanic surface. The bioaccumulation and ingestion of MP debris by the coral polyps have hindered the growth of the corals and forced coral bleaching. Most of the previous studies on MP have focused on the sources and distribution of MP. However, little is known about coral bleaching and its adverse impacts on the coral ecosystem. Therefore, the present review focuses on identifying sources, their global distribution, and the adverse effects of MP on the marine ecosystem, with particular reference to corals and coral reef ecosystems. The current study's findings highlighted that only a few countries and a few researchers worldwide have worked with MP pollution and its life-threatening impact on coral ecosystems. Earlier researchers showed fish communities' vulnerability to MPP in the aquatic environment. Hence, the present work recommends researching MPP and the worldwide threats to coral ecosystems. Apart from this, the production of MP should be minimized after the identification of MP sources, regular monitoring of aquatic ecosystems, recycling of MP elements, and strict government policies to slow down the dreadful impact of MP on coral reef ecosystems and marine environments.
... A marine habitat map that delineates the reef cover and distribution, is useful in determining the reef status and thereby facilitating informed decision-making and effective management . Various initiatives, such as Millennium Coral Reef Mapping Project, Living Oceans Foundation, and Allen Coral Atlas, have employed remote sensing technology, utilizing satellite images to generate marine habitat maps (Andréfouët and Bionaz, 2021;Lyons et al., 2020;Purkis et al., 2019). Despite that, only a quarter of the seafloor has been mapped at a resolution similar to that in terrestrial studies (Wright and Heyman, 2008). ...
... To understand the dynamics of coral reef communities in response to environmental changes, quantifying changes in reef cover using very high-resolution (VHR) satellite images such as LiDAR, Pleiades, Geoeye, Worldview, Quickbird, and IKONOS has allowed us to assess the coral reef degradation and recovery (Ampou et al., 2018;Bajjouk et al., 2019;Lyons et al., 2020). Although VHR data provide detailed and accurate information, they come at a significant cost. ...
... However, existing long-term benthic databases such as Caribbean Coastal Marine Productivity or National Oceanic and Atmospheric Administration lack uniformity, hindering cross-validation, particularly for finer-resolution benthos categorization (Tebbett et al. 2023a(Tebbett et al. , 2023b. Consolidating these datasets under a common protocol would increase the value of data for training and validation of remote sensing mapping algorithms (Lyons et al. 2020). 2. The reliable recording and reporting of survey sites is crucial to data sharing and study replication. ...
Coral reefs are spatially variable ecosystems that form biogenic structures ranging in size from 10 to 1000s of meters. Their changes in response to anthropogenic stress are occurring across increasingly broad scales, yet our ability to detect, understand and respond to these changes at relevant scales is limited. Traditional in-water observation-based coral reef ecology and remote sensing-based methods both offer valuable insights into benthic change, but their relative scalability and use to-date must be understood to inform optimal future research approaches. We conducted a systematic literature review comparing the approaches used to quantify benthic habitat, through traditional in-water ecological studies and remote sensing studies, with respect to: (a) their geographic distribution, (b) reef zone selection, and c) their focal questions. Among the 199 studies reviewed, traditional ecological studies primarily concentrated on community composition (89%), using high-detail direct measurements, especially from the reef slope (80%). By contrast, remote sensing studies provided spatially explicit datasets at coarser spatial and thematic resolutions, with a predominant focus on benthic mapping (72%) across entire reef systems. Only 3% of studies integrated both approaches, combining comprehensive in-situ observations with broadscale remote sensing. As anthropogenic stressors continue to increase in scale, bridging these scientific disciplines offers a promising way to upscale observations to entire reef-scape scales. We identify steps to harness the strengths of both fields and integrate multiple tools at various levels of resolution and scale. Such bridging approaches offer a way forward in understanding and managing coral reef functioning in the Anthropocene.
... Alternatively, in future seascape genomic studies, researchers could record local reef characteristics (e.g., depth and reef habitat type) surrounding each sampled coral. These more fine-scale descriptors could be used to infer local environments that may help explain variation in genotype frequencies and can be easily extracted from the worldwide geomorphic map of the Allen Coral Atlas [69]. In the future, recent advances in photogrammetry may allow for the rapid and affordable modeling of reef 3D structure from underwater videos [70], paving the way for fine-scale genotype-by-environment association analyses that account for topography within a reef [71]. ...
... Drawing information from over 100 trillion pixels acquired by these two earth observation programs, the data stack contained reflectance values, derived reflectance metrics, satellite-derived water depth (primarily from Sentinel-2 imagery 20 ), and modeled wave environment. 21 Depth and waves are non-spectral variables that influence geomorphic zones and benthic substrate composition 21,22 and are thus highly informative resources for determining the distribution of reefs globally. We combined 21 the global data stack with a globally comprehensive training and validation database for the geomorphic (n = 1 million samples) and benthic (n = 600,000 samples) mapping classes. ...
... Drawing information from over 100 trillion pixels acquired by these two earth observation programs, the data stack contained reflectance values, derived reflectance metrics, satellite-derived water depth (primarily from Sentinel-2 imagery 20 ), and modeled wave environment. 21 Depth and waves are non-spectral variables that influence geomorphic zones and benthic substrate composition 21,22 and are thus highly informative resources for determining the distribution of reefs globally. We combined 21 the global data stack with a globally comprehensive training and validation database for the geomorphic (n = 1 million samples) and benthic (n = 600,000 samples) mapping classes. ...
... Other environmental and textural variables (such as slope, wave climate, gray level cooccurrence metrics, band ratios) were calculated and added to the covariate stack. 21 All input data layers were segmented into image ''objects'' as per previously described methods. 21 Training and validation data The training data used were points sampled from a set of reference data (polygons), which were developed specifically for each region via a standardized global protocol. ...
... To conclude, we recommend mixing several input data to improve accuracy: photo transects, underwater camera videos, bathymetry, salinity or temperature measurements [32], [138], [170], [338], [352]. ...
... the Millennium Coral Reef Mapping Project [13], the Allen Coral Atlas [32] or the Khaled bin Sultan Living Ocean Foundation [57]. These maps are proven useful to the scientific community for coral reef and biodiversity monitoring and modeling, as well as inventories or socio-economic studies [14]. ...
... Our work was thought so that the proposed tool is adaptable and can be retrained to handle different types of data. For instance, the workflow has been tested with geomorphic and benthic data of the Great Barrier Reef from the Allen Coral Atlas project [32], by considering their maps as expert-based maps. ...
The ongoing crisis of climate change necessitates the development of effective methods for monitoring and mapping environmental features and species to ensure their preservation. This thesis explores the application of machine learning algorithms to efficiently map coral reefs using multispectral satellite images. The Maupiti lagoon in French Polynesia serves as a case study. The research led to the production of an automated tool capable of generating coral reef maps from satellite images. Moreover, the tool can be adapted to map other ecosystems, such as forests or ice sheets, provided that the model is retrained with relevant data.
To begin, a comprehensive literature review investigates current methods and trends in utilizing machine learning algorithms for coral reef mapping. Then, the attempts to develop the tool led us to face the special case of compositional data, which are data carrying relative information and lying in a mathematical space known as simplex. Adaptations of conventional methods are required to address the specific characteristics of this space.
First, in response to data imbalance, an oversampling technique is developed specifically for compositional data. Additionally, a spatial autoregressive model based on the Dirichlet distribution is formulated to account for spatial effects that may arise in the mapping process.
Finally, we present the implementation of our final mapping tool. To achieve the desired objective, a two-staged classification process is implemented, combining pixel-based and object-based approaches. This technique enables the tool to achieve an accuracy exceeding 85% with 15 classes.
The research contributes novel solutions for handling compositional data and delivers a high-performing mapping tool for coral reef ecosystems, aiding in environmental management and conservation efforts.
