Coastline Detection Research Articles

Event-driven nearshore and shoreline coastline detection on SpiNNaker neuromorphic hardware

Abstract Coastline detection is vital for coastal management, involving frequent observation and assessment to understand coastal dynamics and inform decisions on environmental protection. 
Continuous streaming of high-resolution images demands robust data processing and storage solutions to manage large datasets efficiently, posing challenges that require innovative solutions for real-time analysis and meaningful insights extraction.
This work leverages low-latency event-based vision sensors coupled with neuromorphic hardware in an attempt to decrease a two-fold challenge, reducing the computational burden to ~0.375 mW whilst obtaining a coastline detection map in as little as 20 ms.
The proposed Spiking Neural Network (SNN) runs on the SpiNNaker neuromorphic platform using a total of 18040 neurons reaching 98.33% accuracy.
The model has been characterised and evaluated by computing the accuracy of Intersection over Union (IoU) scores over the ground truth of a real-world coastline dataset across different time windows. The system's robustness was further assessed by evaluating its ability to avoid coastline detection in non-coastline profiles and funny shapes, achieving a success rate of 97.3%.

Semantic segmentation of coastal aerial/satellite images using deep learning techniques: An application to coastline detection

A new CNN based approach supported by semantic segmentation, was proposed. This approach is frequently used to carry out regional-scale studies. The core of our method revolves around a CNN model, based on the famous U-Net architecture. Its purpose is to identify different classes of pixels on satellite images and later to automatically detect the coastline. The recently launched Coast Train dataset was used to train the CNN model. Traditional coastline detection was improved (“water/land” segmentation) by means of two new aspects the use of the Sobel-edge loss function and the segmentation of the satellite images into several categories like built-up areas, vegetation and land besides beach/sand and water classes. The approach used ensures a more precise coastline extraction, distinguishing water pixels from all other categories. Our model adeptly identifies features, such as cliff vegetation or coastal roads, that some models might overlook. In this way, coastline localization and its drawing for regional scale study, have minor uncertainties. The performance of the CNN-based method, achieving 85% accuracy and 80% IoU (Intersection over Union) in the segmentation process. The ability of the model to extract the coastline was validated on a Sicilian case study, notably the San Leone beach (Agrigento). The model's results align closely with the ground truth, moreover, its reliability was further confirmed when it was tested on other Sicilian coastal regions.Beyond robustness, the model offers a promising avenue for enhanced coastal analysis potentially applicable to coastal planning and management.

A novel sea-land segmentation network for enhanced coastline extraction using satellite remote sensing images

The extraction of coastlines from remote sensing images is vital for promoting sustainable development in coastal areas, conserving marine environments, strengthening disaster response capabilities, and supporting scientific research. However, current coastline detection approaches using remote sensing face challenges related to resolution, terrain, boundary, and data, requiring accurate solutions for reliability. Here, we introduce the Collaborative Supervision and Attention Fusion (CSAFNet) model for pixel-level sea-land segmentation, with a primary goal of improving the accuracy of coastline extraction. The model integrates the Edge Deep Supervision (EDS) module to enhance coastline edge detail fitting. Additionally, the Collaborative Semantic Deep Supervision (CSDS) module and Attention Fusion Module (AFM) collaborate to bridge the semantic gap between different hierarchical features, resulting in a more precise and detailed delineation of coastlines. Experimental validation on the publicly available SLSD 11The dataset was initially labeled as ”sea-land segmentation data” by the authors. For ease of discussion, we will adopt the practice of using the initial letters of each word as an abbreviation. dataset has demonstrated superiority over various advanced methods, with an impressive mIoU value of 96.72%. Through simple optimization, detailed and rich coastlines can be extracted, validating the feasibility of coastline extraction.

The influence of spatial resolution on coastline detection by means of multisource remote sensing data

Identifying an instantaneous coastline through remote sensing data poses a challenge due to various factors, such as the intricate nature of the coast and the selection of the appropriate sensor with its inherent spatial resolution. Even so, accurate identification of the coastline through remote sensing data has immense potential to provide reliable and timely information about the coastal environment. The purpose of this study was to evaluate whether there are spatial differences when detecting a coastline by means of multiple remote sensors. We vectorized orthomosaics from UAV flights during high, mean, and low tides manually, creating three reference baselines. The cross-shore distances from the three vectors were compared to the coastline vectors extracted from a water index, using six different satellite data collected on the nearest possible date to that of the UAV flights, with MODIS Terra, Landsat 8, Sentinel-2, RapidEye, PlanetScope, and Google Earth. The results indicate that the cross-shore distance of the coastline vectors depended directly on the spatial resolution of the sensors. Concerning the mean tide, Google Earth provided the highest accuracy, with a marginal error of 6.7 m, while MODIS provided the largest difference with a margin of 449 m. Our results indicate that future satellite-derived coastline detection approaches should consider the relationship between tide amplitude and spatial resolution on sandy shores.

Correction: Review of Segmentation Methods for Coastline Detection in SAR Images

Correction: Review of Segmentation Methods for Coastline Detection in SAR Images

Coastline Detection in Polarimetric SAR Images Based on Freeman Decomposition and Three-Region Markov Random Field Segmentation

Coastline Detection in Polarimetric SAR Images Based on Freeman Decomposition and Three-Region Markov Random Field Segmentation

A Network for Merging SAR Image Sea-Land Segmentation and Coastline Detection Tasks

A Network for Merging SAR Image Sea-Land Segmentation and Coastline Detection Tasks

High-Resolution Real-Time Coastline Detection Using GNSS RTK, Optical, and Thermal SfM Photogrammetric Data in the Po River Delta, Italy

High-resolution coastline detection and monitoring are challenging on a global scale, especially in flat areas where natural events, sea level rise, and anthropic activities constantly modify the coastal environment. While the coastline related to the 0-level contour line can be extracted from accurate Digital Terrain Models (DTMs), the detection of the real-time, instantaneous coastline, especially at low tide, is a challenge that warrants further study and evaluation. In order to investigate an efficient combination of methods that allows to contribute to the knowledge in this field, this work uses topographic total station measurements, Global Navigation Satellite System Real-Time Kinematic (GNSS RTK) technique, and the Structure from Motion (SfM) approach (using a low-cost drone equipped with optical and thermal cameras). All the data were acquired at the beginning of 2022 and refer to the areas of Boccasette and Barricata, in the Po River Delta (Northeastern of Italy). The real-time coastline obtained from the GNSS data was validated using the topographic total station measurements; the correspondent polylines obtained from the photogrammetric data (using both automatic extraction and manual restitutions by visual inspection of orhophotos) were compared with the GNSS data to evaluate the performances of the different techniques. The results provided good agreement between the real-time coastlines obtained from different approaches. However, using the optical images, the accuracy was strictly connected with the radiometric changes in the photos and using thermal images, both manual and automatic polylines provided differences in the order of 1–2 m. Multi-temporal comparison of the 0-level coastline with those obtained from a LiDAR survey performed in 2018 provided the detection of the erosion and accretion areas in the period 2018–2022. The investigation on the two case studies showed a better accuracy of the GNSS RTK method in the real-time coastline detection. It can be considered as reliable ground-truth reference for the evaluation of the photogrammetric coastlines. While GNSS RTK proved to be more productive and efficient, optical and thermal SfM provided better results in terms of morphological completeness of the data.

Coastline detection in satellite imagery: A deep learning approach on new benchmark data

Detailed and up-to-date coastline morphology data underpins our understanding of coastline change over time. The development of an automated and scalable coastline extraction methodology from satellite imagery is currently limited by the low availability of open, globally distributed and diverse labelled data with which to develop and benchmark techniques. Therefore, in this study we present the Sentinel-2 Water Edges Dataset (SWED), a new and bespoke labelled image dataset for the development and bench-marking of techniques for the automated extraction of coastline morphology data from Sentinel-2 images. Composed of 16 labelled training Sentinel-2 scenes, and 98 test label-image pairs, SWED is globally distributed and contains examples of many different coastline types and natural and anthropogenic coastline features.To provide a baseline of model performance against SWED we train and test four convolutional neural network models, based on the U-Net model architecture. Models are optimised using Categorical Cross-entropy Loss, Sørensen–Dice Loss and two novel loss functions we present for the focusing of model training attention to the boundary between land and water. Through a hybrid quantitative and qualitative model assessment process we demonstrate that the model trained using our novel Sobel-edge loss function has greater sensitivity to fine-scale, narrow coastline features whilst possessing near top quantitative performance demonstrated by Categorical Cross-entropy.The SWED dataset is published openly for use by the remote sensing and machine learning communities, whilst the Sobel-edge loss is available for use in machine learning applications where sensitivity to boundary features is important.

Coastline Extraction from GF-3 SAR Images Using LKDACM and GMM Algorithms

Coastline detection using a Gaussian Mixture Model (GMM) applied to synthetic aperture radar (SAR) imagery is usually inaccurate due to the inherent noise of SAR data. In addition, the traditional active counter model is sensitive to the initial position of the contour line and requires a large number of iterations to converge to a solution. In this study, we first used the GMM algorithm to segment the SAR images and obtain a coarse land and sea segmentation map. This map is then used as the initial position for a subsequent active contour model. The K distribution was introduced into the local statistical active contour model to better model the SAR image. The Gaussian distribution-based local active contour model and the algorithm detailed in this paper were used to perform coastline extraction experiments on four SAR images. Four GF-3 SAR images with different modes were collected to validate the efficiency of the proposed method. The experimental results show that the coastline extraction methods from SAR images based on the GMM algorithm and the K distribution-based local statistical active contour model (LKDACM) overcame the shortcomings of the traditional active contour model to accurately and quickly detect coastlines, thus enabling the detection of coastline changes.

Coastline Detection Based on Sentinel-1 Time Series for Ship- and Flood-Monitoring Applications

This letter addresses the use of the Sentinel-1 time series with the aim of proposing an automatic and unsupervised coastline detection method that averages the dynamical variations of coastal areas over a limited period of time, e.g., one year. First, we propose applying a temporal averaging filter that allows the temporal variations in coastal areas, e.g., due to tides or vegetation, to be encapsulated, and, at the same time, the speckle to be reduced, without decreasing the spatial resolution of the synthetic aperture radar (SAR) time series. Then, based on the distinctive backscattering values of the sea and land pixels, we will employ an iterative hierarchical tiling method in order to accurately characterize the two classes using bimodal distribution. The distribution is then segmented by a thresholding and region-growing procedure to separate the sea and land classes. A large-scale quantitative comparison between the SAR-derived and open street map (OSM) coastlines allows for a numerical evaluation of the results, i.e., an overall agreement ranging from 80% to 90%. In addition, Sentinel-2 images are used to evaluate the estimated SAR coastline qualitatively. Furthermore, the benefits of having an accurate SAR coastline are shown in the case of two well-known Earth observation-monitoring applications, ship detection, and floodwater mapping.

HED-UNet: Combined Segmentation and Edge Detection for Monitoring the Antarctic Coastline

Deep learning-based coastline detection algorithms have begun to outshine traditional statistical methods in recent years. However, they are usually trained only as single-purpose models to either segment land and water or delineate the coastline. In contrast to this, a human annotator will usually keep a mental map of both segmentation and delineation when performing manual coastline detection. To take into account this task duality, we therefore devise a new model to unite these two approaches in a deep learning model. By taking inspiration from the main building blocks of a semantic segmentation framework (UNet) and an edge detection framework (HED), both tasks are combined in a natural way. Training is made efficient by employing deep supervision on side predictions at multiple resolutions. Finally, a hierarchical attention mechanism is introduced to adaptively merge these multiscale predictions into the final model output. The advantages of this approach over other traditional and deep learning-based methods for coastline detection are demonstrated on a dataset of Sentinel-1 imagery covering parts of the Antarctic coast, where coastline detection is notoriously difficult. An implementation of our method is available at \url{https://github.com/khdlr/HED-UNet}.

Geospatial Analysis of Coastline Erosion Along Pulau Tuba, Langkawi

The erosion in Malaysia has brought attention to many authorities especially the coastline in the eastern part of Peninsular Malaysia. Although the erosion in the northern part of Peninsular Malaysia does not receive as much attention as the eastern part of Peninsular Malaysia, however, the issue should not be neglected. High spatial resolution satellite imageries were used for the extraction of coastline and classification level of erosion rate along with the Pulau Tuba. The coastline data was extracted using two different methods known as Maximum Likelihood (ML) and On-Screen Digitizing (OSD) in the determination of the best approach of coastline detection from the Sentinel-2 data of the year 2016 and 2019. Furthermore, the level of erosion is made based on the physical and economic parameters outlined by the National Coastal Erosion Study 2015 (NCES). Due to some inevitable constraints of Movement Control Order by the Malaysian government due to the COVID-19 pandemic, physical observation data of Pulau Tuba were collected via Google Maps. The information acquired includes type of coastal geomorphology, land use, development on the area, activities conducted, and adaptation of erosion if any. These data were utilized to determine the erosion rate and categories using the proposed model by NCES for five divided management units (MU) of the Pulau Tuba areas utilizing Erdas Imagine and ArcGIS software. The analysis found that the ML approach has under-detected the coastline length between 3.19% to 45.0% as compared to OSD for both years of 2016 and 2019. Rate of erosion for Pulau Tuba based on the NCES approach found that the highest erosion rate occurred at the MU1 (Pulau Dayang Bunting- Pulau Tuba causeway) with 2.91% and classified as K1 (critical erosion category) with a value of 4.39 m/yr−1 and the highest accretion rate at the MU3 with 3.06%. The critical erosion category was associated with the MU that has significant development and on-going activities that occurred in the area especially in MU 4 (Pulau Tuba) and MU 5 (Teluk Berembang). Other than that, the high number of erosions occurred in that section is due to the exposure of waves, wind, currents, and tides.

Remote Sensing in Coastline Detection

“Is beach erosion a natural cycle or is it getting worse with rising sea levels [...]

Coastline detection in polarimetric SAR images using Markov random field segmentation based on mixture Wishart distribution

Coastline detection in polarimetric SAR images using Markov random field segmentation based on mixture Wishart distribution

Knowledge technologies based on fabrication process composite materials and remote sensing applications

The aim of this work concentrates on utilizing powerful MATLAB programming (software version R2016a) to evaluate the impact of environmental variations of water case in the Mosul Dam reservoir and observed its receding impact on human life activities based on composite image processing applications. Furthermore, composite materials of different temporal remote sensing data increase powerfully the estimation of environmental variables of relevance to human health. Thus, temporal remote sensing data trends to enhance the efficiency of detecting receding water resources effect of human life impacts over different years. Two steps were implemented, which focuses on the estimation of changes in the water surface of the lake over 31 years. Preprocessing step concentrates on composite data materials from different Landsats to be more suitable for next step by utilizing color composite image processing and postprocessing step implemented the coastline detection of the reservoir and recognition of the quality of clear water in the lake due to the variation of water spectral reflectivity by hybrid classification method. The performance of this study is based on statistics measurements on the surface area of water level and overall accuracy, which indicated that hybrid classification method improves the capacity of integrating two classification methods, which gained highly identification water lake classes regarding its quality and more. The obtained results achieved the desired purpose of this study to investigate the high power application through implementing composite different image processing techniques with temporal satellite data to conversance the amount of water level changes in Mosul Dam reservoir and its impact on storage quantity over years.

ОБНАРУЖЕНИЕ БЕРЕГОВОЙ ЛИНИИ ПО ДАННЫМ ДЗЗ СРЕДНЕГО ПРОСТРАНСТВЕННОГО РАЗРЕШЕНИЯ

Remote sensing data are widely used in coastal zones monitoring, since they provide radiometric information with the possibility to automate the data processing. Due to the lack of high resolution satellite images in free access using the medium resolution satellite images is widespread. The study is dedicated to the development of a coastline detection method based on medium resolution satellite images. It is proposed to use a semi-automatic method based on uncontrolled classification of a mid-resolution image by water indices, followed by expert refinement of classes and the use of machine learning methods. The shorelines of the eastern part of the Gulf of Finland have been extracted from Landsat 8 Level-2 image, using proposed method. The position accuracy of the generated shorelines has been analyzed using manually digitized shoreline from high-resolution image Resurs-P1 processing level 2А. The results showed that in the test areas the best output for extracting the coastline are given by the MNDWI with a fixed threshold value equal to zero. The Random Forest machine learning algorithm was used to refine the type of coastline, especially in wetlands where water indices showed poor accuracy. The study showed a significant increase in the position accuracy with the use of the algorithm. However, the accuracy of manually classified wetlands for training the model has a significant impact on the result.

Hierarchical coastline detection in SAR images based on spectral‐textural features and global–local information

This study presents a novel approach to detect the coastline from single-polarisation synthetic aperture radar (SAR) images. The proposed method encompasses land/sea segmentation, coastline detection, and refinement. A novel spectral–textural segmentation framework (STSF) is proposed by using the spectral–textural features extracted from the input image patches. The STSF distinguishes various coastal/sea types and is robust to noise. Also, a hierarchical region-based level set method (LSM) is proposed to detect the coastline, accurately. The first LSM step applies global information for evolution. The LSM initialisation is performed using the obtained rough segmentation, which is very practical as the final LSM evolution depends on the initial value, particularly on complex SAR images. The global region-based LSM (GRB-LSM) step modifies the previous segmentation and approaches to the coastline. To improve accuracy, a local region-based LSM (LRB-LSM) is proposed. Therefore, in the second LSM step, the LRB-LSM applies to the results of GRB-LSM step. The LRB-LSM improves the accuracy of the detected coastline while ensuring its smoothness. To verify the performance of the proposed method, several high-resolution SAR images from different microwave bands and various coastal environments are used. The performance of the proposed method is confirmed by the given experiments.

Automatic Methodology to Detect the Coastline from Landsat Images with a New Water Index Assessed on Three Different Spanish Mediterranean Deltas

Due to the importance of coastline detection in coastal studies, different methods have been developed in recent decades in accordance with the evolution of measuring techniques such as remote sensing. This work proposes an automatic methodology with new water indexes to detect the coastline from different multispectral Landsat images; the methodology is applied to three Spanish deltas in the Mediterranean Sea. The new water indexes use surface reflectance rather than top-of-atmosphere reflectance from blue and shortwave infrared (SWIR 2) Landsat bands. A total of 621 sets of images were analyzed from three different Landsat sensors with a moderate spatial resolution of 30 m. Our proposal, which was compared to the most commonly used water indexes, showed outstanding performance in automatic detection of the coastline in 96% of the data analyzed, which also reached the minimum value of bias of − 0.91 m and a standard deviation ranging from ±4.7 and ±7.29 m in some cases in contrast to the existing values. Bicubic interpolation was evaluated for a simple sub-pixel analysis to assess its capability in improving the accuracy of coastline extraction. Our methodology represents a step forward in automatic coastline detection that can be applied to micro-tidal coastal sites with different land covers using many multi-sensor satellite images.

Improvement of EPIC/DSCOVR Image Registration by Means of Automatic Coastline Detection

In this work, we address the image geolocation issue that is present in the imagery of EPIC/DSCOVR (Earth Polychromatic Imaging Camera/Deep Space Climate Observatory) Level 1B version 2. To solve it, we develop an algorithm that automatically computes a registration correction consisting of a motion (translation plus rotation) and a radial distortion. The correction parameters are retrieved for every image by means of a regularised non-linear optimisation process, in which the spatial distances between the theoretical and actual locations of chosen features are minimised. The actual features are found along the coastlines automatically by using computer vision techniques. The retrieved correction parameters show a behaviour that is related to the period of DSCOVR orbiting around the Lagrangian point L 1 . With this procedure, the EPIC coastlines are collocated with an accuracy of about 1.5 pixels, thus significantly improving the original registration of about 5 pixels from the imagery of EPIC/DSCOVR Level 1B version 2.