Emerging Techniques in Satellite-Derived Shoreline

Analysis of shoreline variability and shoreline erosion-accretion trends is fundamental to a broad range of investigations undertaken by coastal scientists, coastal engineers, and coastal managers. Though strictly defined as the intersection of water and land surfaces, for practical purposes, the dynamic nature of this boundary and its dependence on the temporal and spatial scale at which it is being considered results in the use of a range of shoreline indicators. These proxies are generally one of two types: either a feature that is visibly discernible in coastal imagery (e.g., highwater line [HWL]) or the intersection of a tidal datum with the coastal profile (e.g., mean high water [MHW]). Recently, a third category of shoreline indicator has begun to be reported in the literature, based on the application of imageprocessing techniques to extract proxy shoreline features from digital coastal images that are not necessarily visible to the human eye.

Earth Science Informatics, 2017

Video systems have become widely used all around the world in coastal monitoring strategies, allowing both high temporal and spatial sampling frequency, with low logistic and costs efforts. The present paper deals with a new tool for coastal images processing, aimed at the automatic shoreline detection and data analysis. The tool is composed by a shoreline detection routine implemented in a web-application, addressed at images processing (i.e. shoreline extraction and geo-rectification), data analysis and sharing results about beach actual state and shore evolution in quasi-real time. The Shoreline Detection Model (SDM) is based on a new algorithm, implementing image-processing procedures, which allows extracting the sea/ land boundary from automatic segmented Timex images. The SDM calibration and validation has been performed on different coastal images derived from a video monitoring system installed at Alimini (Lecce, IT) in 2005, by comparing automatic shoreline contours with the manual detected ones. Moreover, in December 2015, new video monitoring systems were installed in South Italy (Porto Cesareo and Torre Canne, Apulia region), at sandy beaches affected by erosion phenomena. The application of the SDM on images recorded by the new systems has allowed testing the model feasibility at sites characterized by different morphological features and geographical exposition. The present describes in detail the SDM algorithm and the image processing procedures used. The results of the model calibration and validation performed at Alimini and the tests performed at Porto Cesareo on first images are reported.

Shoreline extraction is a key process for many coastal zone applications, such as navigation and coastal environmental protection. The manual extraction of shorelines manually is tedious and subject to the operator's ability. The main objective of this research is to evaluate the use of two different image fusion techniques (IHS and PCA-Principal Component Analisys) using near-infrared band on multispectral Sentinel-2 imagery to extract shoreline in the coastal zone of Cassino beach, Southern Brazil. The resulting images were classified into two classes (water and non-water) using the K-Means algorithm, and the accuracy was evaluated through the analysis of mean absolute difference and RMSE applied on segments of artificial coastal structures. The results indicate that the shoreline extraction by PCA method obtained the most accurate results, and the use of sharpened MNDWI (Modified Normalized Difference Water Index) image shows a good alternative to improve shoreline extraction.

eproceedings.worldscinet.com

Shoreline identification using satellite images is compared with in situ shoreline measurements in the Yucatan Peninsula to evaluate its potential for studying shoreline changes in places with a paucity of data. This study firstly tests the detection limits of shoreline identification by comparing a SPOT image with ground shoreline measurements, and secondly we show examples of overlaying satellite-derived shorelines from three different years to assess the ability of the technique to quantify real shoreline changes. The mean (-0.19 m) and the standard deviation (4 m) between the ground and satellite-derived shoreline are much smaller than the pixel size. Shoreline changes of more than 30 m were measured between images spanning several years (2004, 2006 and 2008) in areas near to coastal structures and near urban areas without coastal vegetation.

In the last 50 years, the inhabitants of the 19 municipalities along the Abruzzo coast have doubled and a stronger impact of the activities connected with the tourism has been experienced. The area, naturally exposed to the effects of changes of the sea level, has been characterized by a dramatic increase of erosion, due to the reduction of the solid transport from rivers to sea, as a consequence of extensive works carried out on the watersheds, to mitigate extreme rainfall and consequent flooding. The availability of data acquired by different sensors on the last few decades might be useful to assess the overall accretion/erosion trend of coastline, whereas combination of different observations taken in a restricted timeframe, may provide interesting inputs for detailed studies. In the present study, a methodology is proposed for the coastline identification from WorldView-2 images, available in 8 spectral bands, with 0.5 m and 1.8 m of spatial resolution for the panchromatic and the multispectral channels respectively. In particular, a pixel based multispectral classification is used to identify various types of land cover. The 8 bands allow to get good results, both in the classification process and with NDVI, NDWI, SAM, FM algorithms, for the identification of various land cover and in particular to separate dry from wet sand. Interesting results were obtained testing an algorithm to evaluate the relative depth of the water using the "coastal blue" band. Interesting results provided the comparison of the estimated coastline with such methodology and a topographic map of same area. This comparison highlights the changes in the study area. There are many possible applications of the proposed techniques, such as map updates and coastal change monitoring.

Analysis of shoreline variability and shoreline erosion-accretion trends is fundamental to a broad range of investigations undertaken by coastal scientists, coastal engineers, and coastal managers. Though strictly defined as the intersection of water and land surfaces, for practical purposes, the dynamic nature of this boundary and its dependence on the temporal and spatial scale at which it is being considered results in the use of a range of shoreline indicators. These proxies are generally one of two types: either a feature that is visibly discernible in coastal imagery (e.g., high-water line [HWL]) or the intersection of a tidal datum with the coastal profile (e.g., mean high water [MHW]). Recently, a third category of shoreline indicator has begun to be reported in the literature, based on the application of image-processing techniques to extract proxy shoreline features from digital coastal images that are not necessarily visible to the human eye. Potential data sources for shoreline investigation include historical photographs, coastal maps and charts, aerial photography, beach surveys, in situ geographic positioning system shorelines, and a range of digital elevation or image data derived from remote sensing platforms. The identification of a " shoreline " involves two stages: the first requires the selection and definition of a shoreline indicator feature, and the second is the detection of the chosen Geo-Spatial Technologies In Shoreline Analysis, Variability and Erosion http://www.iaeme.com/IJCIET/index.asp 77 editor@iaeme.com shoreline feature within the available data source. To date, the most common shoreline detection technique has been subjective visual interpretation. Recent photogrammetry, topographic data collection, and digital image-processing techniques now make it possible for the coastal investigator to use objective shoreline detection methods. The remaining challenge is to improve the quantitative and process-based understanding of these shoreline indicator features and their spatial relationship relative to the physical land–water boundary.

IEEE Data Descriptions, 2025

Shoreline segmentation is a specialized water-land segmentation task that detects the position of the shoreline. Existing datasets for segmentation either do not focus on coastal environments (e.g., DeepGlobe) or are not based on high-resolution imagery (e.g., Sen1Floods11). To fill this gap, we built Coastal Aerial Imagery Dataset (CAID), a high-resolution, manually labeled benchmark dataset built on aerial imagery for shoreline segmentation. CAID consists of 20,689 images of 500 × 500 pixels across 850 NAIP sites across the Great Lakes region. The dataset covers a variety of coastal landscapes and its representativeness of coastal environments was evaluated using three criteria: water body area, shoreline length ratio, and mean water hue. We also assessed the benchmark performance of existing segmentation models on CAID. While these models achieved high mIoU and mAcc scores, accurately positioning shorelines remains a challenge, a limitation that is not fully reflected by existing evaluation metrics. To address this, we introduced mean shoreline covering ratio (mSCR), a novel metric that quantifies each model's ability to align with ground-truth shorelines. The CAID dataset, along with the associated code for training and evaluating deep learning models, will be publicly released under the CC-BY-4.0 license via GitHub and Zenodo to ensure open-access availability.

Automatic Shoreline Detection and Change Detection Analysis of Netravati-GurpurRivermouth Using Histogram Equalization and Adaptive Thresholding Techniques, 2015

The shoreline change extraction and change detection analysis is an important task that has application in different fields such as development of setback planning, hazard zoning, erosion-accretion studies, regional sediment budgets and conceptual or predictive modeling of coastal morphodynamics. Shoreline delineation is difficult, time consuming, and sometimes impossible for entire coastal system when using traditional ground survey techniques. Recent advances in remote sensing and geographical information system (GIS) techniques are overcoming the difficulties in extraction of shoreline position and detection of shoreline changes. In the present paper, an automatic shoreline detection method using histogram equalization and adaptive thresholding techniques is developed. The shoreline of Netravati-Gurpur rivermouth area along Mangalore coast, West Coast of India have been extracted from Indian Remote Sensing Satellite (IRS P6) LISS-III (2005, 2007 and 2010) and IRS R2 LISS-III (2013) satellite images using developed automatic shoreline detection method. The delineated shorelines have been analyzed using Digital Shoreline Analysis System (DSAS), a GIS Software tool for estimation of shoreline change rates through two statistical techniques such as, End Point Rate (EPR) and Linear Regression Rate (LRR). The Bengre spit, Northern sector of Netravati- Gurpur river mouth is under accretion an average of 2.95 m/yr (EPR) and 3.07 m/yr (LRR) and maximum accretion obtained is 8.51 m/yr (EPR) and 8.69 m/yr (LRR). Southern sector, the Ullal spit is under erosion an average of -0.56 m/yr (EPR) and -0.59 m/yr (LRR). © 2015 The Authors. Published by Elsevier B.V. Peer-review under responsibility of organizing committee of ICWRCOE 2015. Keywords:Histogram equalization, thresholding, shoreline, remote sensing, digital shoreline analysis system

Machine Learning (ML) is a discipline of computer science that develops dynamic algorithms capable to produce data-driven decisions (Thessen, 2016). ML has proven itself to be an answer to many real-world problems with its capabilities. ML has the advantage over the traditional methods because it can a build model which is highly dimensional and nonlinear data with complex relations and missing values. ML has proven useful for a very large number of applications in many parts of the Earth system (land, ocean, and atmosphere) and beyond, from retrieval algorithms, crop disease detection, new product creation, bias correction and code acceleration (Yi & Prybutok, 1996). A large amount of data is collected by scientific instruments then separated into train set and test set. Therefore, ML algorithms are trained by this data. then build model with high accuracy and its parameters are optimized based on sample data during the learning step. During prediction the model parameters are used to infer results on the previously unseen data. ML has multiple algorithms, techniques, and methodologies that can be used to build models to solve real-world problems using oceanographic data. (a) A supervised Learning (SL) is a type of ML algorithm that uses labeled data. After that, the machine is provided with new set of data so that SL algorithms analyses the training data and produce a correct outcome from the labeled data. SL mainly trials to model the relationship between the inputs and their corresponding outputs from the training data so that we would be able to predict the output based on the knowledge it gained earlier with regard to relationships. SL are classified into two major categories. a. classification and b. regression. (b) Unsupervised learning (USL) is the training of the machine using data that is neither classified nor labeled. The task of the machine is to group unsorted data based on the similarities, patterns and differences without any guidance. USL can be classified into following the categories a. clustering, b. dimensionality reduction c. anomaly detection. (c) The reinforcement learning (RL) methods are slightly different from SL or USL. RL is a type of ML where an agent learns how to behave in the environment by performing actions and thereby drawing intuitions and seeing the results. (d) Deep learning (DL) is the subset of ML concerned with algorithms inspired by the structure and function of the human brain called an artificial neural network. Neural networks (NNs) come in several forms such as recurrent neural networks, convolutional neural networks, and artificial neural networks and feed forward neural networks. An ANN is an interconnected group of nodes. Here, each circular node represents an artificial neuron and an arrow represents a connection from the output of one artificial neuron to the input of another (Burkitt, 2006).However, in the present circumstances of limited and poor infrastructural opportunities for oceanographic research in Bangladesh, ML techniques could be a robust, fast, and highly accurate option.

International journal of environment and geoinformatics, 2023

Coastal areas constitute the most important part of the world when considered in terms of their socioeconomic and natural values. Measuring and monitoring the coastal areas accurately is an important issue for coastal management. Compared to ground-based studies, remote sensing applications enriched with machine learning algorithms such as Random Forest (RF) and Support Vector Machine (SVM) provide significant benefits in terms of cost, time, and size of the study area. Within the scope of this study, Sentinel-2 images for five coastal areas located in Turkey with different morphological and hydrodynamic properties were classified as land and water-bodies using SVM and RF algorithms. Water-body segmentation results of the SVM and RF classification for the different band combinations of Sentinel-2 images have been compared. The reasons affecting the results of the accuracy analysis were examined in accordance with the geography of each area. The accuracy analysis (Accuracy, IoU, F1 Score, Precision, Recall) results for all study areas, which consider different classification methods and band combinations are in the order of 99%. Among the band combinations, the best results were obtained in combinations of BRN for İzmir Aliağa and Hatay Samandağ regions, BGRN for Sakarya Karasu region, GRN for Rize Iyidere region and BGN for Samsun Bafra region. In addition, experimental results show that the utilized machine learning methods provide satisfactory results for combinations involving the NIR band in all study areas.