Figure - available from: GIScience & Remote Sensing
This content is subject to copyright. Terms and conditions apply.
Study region is located on the South-West coast of Florida in the United States as indicated by the red rectangle in the inset map. It expands from Desoto (North Orange marker) to Marco Island (South Orange marker) with Boca Grande (Middle Orange marker) in the mid-point. Based on the availability of ground-truth data and the tile position of Sentinel-2 imagery, it was divided into four main areas of interest (i.e. AOI 1–4).
Source publication
This study examines the use of the Multi-Spectral Instrument (MSI) in Sentinel-2 satellite in combination with regression-based random forest models to estimate bathymetry along the extended southwestern Florida nearshore region. In this study, we focused on the development of a framework leading to a generalized Satellite-Derived Bathymetry (SDB)...
Similar publications
Ocean–land waveform classification (OLWC) is crucial in airborne LiDAR bathymetry (ALB) data processing and can be used for ocean–land discrimination and waterline extraction. However, the accuracy of OLWC for single-wavelength ALB systems is low given the nature of the green laser waveform in complex environments. Thus, in this study, a deep learn...
Citations
... The ML methods have the highest accuracy for L1C data (Fig. 8). In fact, ML-based satellite bathymetric techniques have previously been used in nearshore shallow waters or large inland lakes [41], [42], [43], [44], [45]. In these previous studies, ML-based methods also yielded higher accuracies than either traditional regression or physicsbased methods did. ...
Supraglacial lakes play an important role in the surface mass balance of ice sheets. With global warming, supraglacial lakes may become more extensive on ice sheet surfaces than they currently are. Therefore, accurate estimation of the volume of supraglacial lakes is important for characterizing their impact on ice sheets. In this study, we present a machine learning-based method for estimating the depth of supraglacial lakes through the combination of ICESat-2 ATL03 data with multispectral imagery. We tested this method via Landsat-8 and Sentinel-2 imagery and evaluated the accuracy of the algorithm on 7 test lakes on the Greenland Ice Sheet. Our results show that machine learning-based algorithms achieve better accuracy than traditional regression or physics-based methods do, especially for deeper lakes. The best accuracy was achieved when extreme gradient boosting was applied to a Sentinel-2 L1C image, with root mean square error, mean absolute error, and median absolute error values of 0.54 m, 0.43 m, and 0.36 m, respectively. Furthermore, we evaluated the effects of atmospheric corrections of multispectral imagery in the retrieval of supraglacial lake depth. On the basis of our results, we recommend the direct use of top-of-atmosphere reflectance products in mapping supraglacial lake bathymetry because of the low performance of atmospheric corrections for water and snow/ice in both the Landsat-8 and Sentinel-2 datasets. This study is expected to provide a more efficient method for estimating the depth of supraglacial lakes and laying the foundation for accurately quantifying meltwater volumes over large surface areas in subsequent studies.
... An alternative for water depth detection in shallow waters (<25 m), where multi-beam surveys are often limited, is airborne light detection and ranging (LiDAR). The efficacy of LiDAR depends on factors such as turbidity and wave action (Mudiyanselage et al. 2022). Despite its usefulness, LiDAR has logistical constraints during deployment and limitations associated with transient water turbidity and breaking waves, which can restrict its spatial coverage (Hamylton 2017). ...
... In recent decades, remotely sensed technology has emerged as a low-cost, time-efficient, and widely adopted solution for SDB, presenting a promising alternative tool for mapping bottom depths in areas with highly dynamic seabed characteristics (Casal et al. 2019;Mateo-Pérez et al. 2020;Mudiyanselage et al. 2022). SDB from optical space-borne sensors are often derived based on the radiative transfer of light in water as a link to water depth, which is influenced by the characteristics of seawater (Al Najar et al. 2021;Lumban-Gaol, Ohori, and Peters 2021;Wu et al. 2022). ...
... Most SDB studies traditionally focus on transparent or low-turbidity environments to minimize the impeding effect of turbidity in nearshore areas. Nevertheless, achieving effectiveness and accuracy with such SDB methods in high-turbidity and/or highwave waters remain challenging (Caballero and Stumpf 2019;Mudiyanselage et al. 2022). ...
... However, these approaches were characterized by their time-consuming nature, labor-intensive requirements, high costs, and limited coverage areas (Lu et al. 2019). The remote sensing water depth monitoring technique gained widespread adoption due to its real-time capabilities, extensive coverage, and cost-effectiveness when compared to traditional methods (Zhou et al. 2023;Mudiyanselage et al. 2022;Najar et al. 2022;Yang et al. 2022a, b;Liu and Yue 2017). Therefore, optical satellites have emerged as crucial data sources for water depth inversion. ...
Accurately mapping lake water depth is the fundamental basis for comprehending variations in lake water levels and storage, thereby playing a crucial role in assessing the lake ecosystem and hydrological cycle. This study took Hongjiannao Lake as the research area. Multi-spectral data of 7 experimental areas near the shore of the lake were obtained by unmanned aerial vehicle (UAV), and a water level recorder was used to collect water depth data of the lake from April 20 to May 10, 2023. The depths of shallow and deep waters in the lakes were inverted using two linear models, namely multiple linear regression (MLR) and partial least squares regression (PLSR), as well as six nonlinear models, including BP neural network regression (BP), support vector regression (SVM), random forest regression (RF), extreme gradient boosting (XGBOOST), deep neural network (DNN), and convolutional neural network (CNN). The UAV inversion results at a resolution of 0.35 m were resampled to 10 m and used as field-measured shallow water depth data. Based on Sentinel-2 on April 18, 2023, a shallow water depth inversion model was developed. The measured water depth data combined with Sentinel-2 were used to estimate the depths of deep waters, and the performance of the inversion model was evaluated using the coefficient of determination (R²), root mean square error (RMSE), mean absolute error (MAE), and the ratio of performance to deviation (RPD). The results showed that the accuracy of water depth inversion from the nonlinear model far surpassed that attained by the linear model. The DNN model was the best in shallow water depth inversion (R²=0.87, RMSE = 0.19, MAE = 0.11, RPD = 2.78). Compared to other models, the DNN model was particularly suitable for handling datasets with a large volume of data (N = 10320) and high nonlinearity. It proved that the SVM model was the optimal choice for deep water depth inversion (R²=0.89, RMSE = 0.39, MAE = 0.29, RPD = 3.06), exhibiting significant advantages in processing small sample data (N = 136). The maximum water depth was 5.2 m, and the average was 3.87 m on April 18, 2023. About 67% of the lake water depth ranged from 3.8 m to 5.2 m, about 17% ranged from 2.2 m to 3.8 m, and the water depth was less than 2.2 m, accounting for 16%. Overall, Hongjiannao Lake is a shallow lake. The water depth of inland lakes can be accurately determined based on UAV and Sentinel-2 data, enabling effective monitoring of dynamic changes in lakes and facilitating the formulation of lake protection policies.
... In recent years, research on empirical SDB models has put more attention on introducing Machine Learning (ML). The ML-based SDB model builds a feature extraction and mapping function, such as Support Vector Regression, Multi-Layer Perceptron and Random Forest (Ceyhun and Yalçın, 2010;Misra et al., 2018;Kaloop et al., 2021;Mudiyanselage et al., 2022;Wu et al., 2022), whose inputs come from the reflectance of different bands of multispectral imagery and uses in-situ data to complete the fitting. The increased complexity in feature mapping of the ML models facilitates the establishment of fitting functions that are far more complex than simple linear regression, resulting in the inversion accuracy of these ML-based SDB models outperforming that of traditional algorithms. ...
... One of the intended applications of SDB is to achieve rapid bathymetry inversion in inaccessible regions that lack in-situ water depth measurements, which implies that the generalization ability of the SDB model is likely the most pivotal factor after accuracy. For semi-empirical and empirical SDB models that rely on in-situ data, only a small number of studies have considered the geographical independence of the training and testing datasets or evaluated the transferability (Viaña-Borja et al., 2023;Lumban-Gaol et al., 2022;Mandlburger et al., 2021;Mudiyanselage et al., 2022), yet these factors are crucial for evaluating the generalizability of the SDB models. Alternations in geographical locations result in changes in various physical parameters related to the water column and bottom type, necessitating the assessment of the SDB model's generalizability over a wider range. ...
... Furthermore, the SMART-SDB employs the K-NN approach to amalgamate multiple PLBR models corresponding to different band combinations for inverting river channel bathymetry, and, as its paper says, it is also suitable for nearshore shallow water bathymetry inversion (Niroumand-Jadidi et al., 2020). With respect to ML models, which have garnered notable attention in recent SDB research, this study uses ML-based SDB models including Random Forest Regressor (RFR) (Poursanidis et al., 2019;Wu et al., 2022;Mudiyanselage et al., 2022) and CatBoost (Lowell and Rzhanov, 2024). While the LBR and PLBR are based on linear regressions, the time cost for training is extremely fast and only approximately 2 s. ...
... If depth is often retrieved in fluvial systems by assuming a linear relationship with the reflectance measured by the sensor, in coastal environments machine learning models and nonlinear regressions are becoming more and more common (Alevizos 2020;Marcello et al. 2018;Mudiyanselage et al. 2022;Wei et al. 2021;Zhou et al. 2023). These approaches could be promising for depth retrieval in fluvial environments: a recent study conducted using an experimental channel found that a random forest approach was better suited for retrieving bathymetry under optically complex conditions (e.g. ...
... One such improvement is the use of multi-scene approaches rather than single-image approaches. Although some authors have applied different methodologies to obtain SDB from a single satellite image (Babbel et al., 2021;Quilleuc et al., 2021;Niroumand-Jadidi et al., 2020), it has been demonstrated that a multi-scene approach retrieves depths with higher accuracy in diverse sites compared to single image approaches (Lubac et al., 2022;Mudiyanselage et al., 2022;Traganos et al., 2018;Xu et al., 2021). In this study, we Fig. 10. ...
Coastal zones are strategic environments of high socioeconomic, political, and ecological value, with over half of the world's population residing within 200 km of the coast. This proximity highlights their vulnerability to extreme events, which are exacerbated by global changes, leading to significant coastal impacts such as erosion, flooding, and ecosystem services deterioration. Consequently, efficient and operational methodologies for continuous monitoring are urgently needed to face these challenges. Bathymetric data are essential for understanding coastal dynamics, yet traditional data collection methods are often constrained by logistical challenges and high costs. Spaceborne remote sensing techniques offer significant advantages over traditional ground-based methods, particularly in terms of cost-effectiveness and operational efficiency. Over the last half-century, different Satellite-derived bathymetry (SDB) methodologies have been developed; however, challenges still persist. In this research, we applied a robust SDB methodology to three different study sites: Cíes Islands, Baiona Bay, and Vao beach within the Ría de Vigo, Galicia (NW Spain). These areas offer diverse and complex mesotidal environments to test for the very first time the methodology's efficacy. SDB was retrieved with a median absolute error (MedAE) ranging from 0.35 m to 1.55 m for depths up to 14 m. Results with different data source were evaluated, obtaining MedAE for nautical charts ranging from 0.46 m to 1.55 m. The precision between the data sources were quite close. In addition, multi-image composite was generated using images coinciding with both low tide (LT) and high tide (HT) conditions across the three zones. The lowest MedAE values were consistently obtained in images classified as LT (0.46 m) corresponding to Vao area. The results highlight the potential of nautical charts as a reliable source of calibration data for SDB, confirm the effectiveness of multi-image and switching models to correct artifacts and turbidity, considering tidal effects, improving single image approaches, and leverage visible bands for precise depth retrieval under varying conditions.
... This technique provides low-cost, high-efficiency bathymetric surveys in shallow waters, meeting the requirements for short cycles and timeliness. [1][2][3][5][6][7][8][9] Water depth is estimated by SDB technology based on correlations between the remotely sensed reflectance values of satellite imagery observed with optical multispectral sensors and the water depth during image acquisition. While SDB can be generally applied to depths up to 20 m, it may only be applicable to depths of up to 10 m, depending on the characteristics of the marine region. ...
... 3 Accordingly, recent research has actively focused on developing depth estimation models using various machine learning algorithms and independent satellite imagery to ensure universality. [1][2][3]5,11,15 Random forest (RF) is a machine learning algorithm that falls under the category of decision tree learning. It is commonly employed for tasks involving classification and regression analysis. ...
... 15,[18][19][20][21][22][23][24][25][26] Most SDB research has been conducted in marine waters, where the concentration of suspended solids is low, and the concentration of phytoplankton is less than an annual average of 1.0 mg/m 3 , the underwater transparency is extremely high, or in atolls and coastal waters with low turbidity. 2,3,5,[9][10][11][27][28][29] The coastal waters of the Korean Peninsula's West, South, and East Seas differ considerably in marine environmental characteristics, including depth distribution, water turbidity, and sediment composition. The Yellow Sea (West Sea) seabed comprises sand and mud and is characterized by continuous sediment influx from rivers, seabed topography with low-gradient slopes, extensive tidal flats due to a large tidal range, shallower depths, strong tidal influence, and high underwater turbidity due to consistent tidal currents. ...
... (Traganos et al. 2018). FurthermoreLi et al. (2021) uses clean water Mosaic with minimal water column attenuation, thereby enabling automatic bathymetry estimation algorithms to reduce uncertainties caused by water column attenuation.Mudiyanselage et al. (2022) also utilizes machine learning algorithms through a cloud-based random forest approach to extract shallow water bathymetry information. ...
Mapping coastal areas generally requires large data constellations in time series and requires analysis using complex mathematical and modeling approaches. In shallow-water bathymetric mapping, remote sensing plays an important role in supporting conventional bathymetric mapping, especially in areas that are difficult to access. This method called Satellite Derived Bathymetry (SDB). The cloud computing approach is a solution for mapping shallow water bathymetry rapid and effectively. This study using Google Earth Engine (GEE) to compute remote sensing data for produce near-shore bathymetry. The method of Li et al. (2021) performs bathymetric extraction without using depth samples but uses chlorophyll-A as input for depth extraction parameter calculations. This study examines a small bay in the waters of Pacitan, Anakan Bay, and the waters of Kemujan Island in the Karimunjawa Islands. Within this study area, significant differences in resulting depth are very limited, ranging from 0 to -17.8. The developed model, based on the algorithm proposed by Li et al. (2021), is estimated to be able to provide accurate predictions of up to around 90% in the waters studied, with a root mean error rate (RMSE) of 1.1 meters.
... Accuracy is still a "Pandora's box" that continues to haunt the results of SDB processing, and is an issue that continues to be studied in depth and reported on an ongoing basis [9], depending on variations in the quality of satellite imagery [17], the quality and content of aquatic suspensions [18], and mapping methods [19]. Two general approaches used to obtain shallow water depth via satellite include physicsbased with the assumption of a radiation transfer model [17] and [20], and empirical algorithm-based developed by [21] and [22] combined with multiple regression and log spectral ratio regression [17]. ...
... Accuracy is still a "Pandora's box" that continues to haunt the results of SDB processing, and is an issue that continues to be studied in depth and reported on an ongoing basis [9], depending on variations in the quality of satellite imagery [17], the quality and content of aquatic suspensions [18], and mapping methods [19]. Two general approaches used to obtain shallow water depth via satellite include physicsbased with the assumption of a radiation transfer model [17] and [20], and empirical algorithm-based developed by [21] and [22] combined with multiple regression and log spectral ratio regression [17]. Several recent investigations have shown good results in the accuracy of water depth estimation using machine learning (ML) approaches [23] with RMSE below 1 meter [8] and [24]. ...
... Accuracy is still a "Pandora's box" that continues to haunt the results of SDB processing, and is an issue that continues to be studied in depth and reported on an ongoing basis [9], depending on variations in the quality of satellite imagery [17], the quality and content of aquatic suspensions [18], and mapping methods [19]. Two general approaches used to obtain shallow water depth via satellite include physicsbased with the assumption of a radiation transfer model [17] and [20], and empirical algorithm-based developed by [21] and [22] combined with multiple regression and log spectral ratio regression [17]. Several recent investigations have shown good results in the accuracy of water depth estimation using machine learning (ML) approaches [23] with RMSE below 1 meter [8] and [24]. ...
Shallow water bathymetric information is important for human life because it has a strong influence on phenomena and dynamics in coastal areas. Conventional bathymetric mapping methods are capable of obtaining high-precision accuracy, but require expensive and complex resources. Specifically, in shallow waters, survey instruments have difficulty obtaining adequate depth data due to the many obstacles that must be overcome. Remote sensing-based shallow water bathymetry called as Satellite-Derived Bathymetry (SDB) is a reasonable and efficient choice. Technological developments enable SDB data processing to be much more efficient in terms of time and storage by utilizing cloud-based platforms such as Google Earth Engine (GEE). Spatial and temporal resolution is still a challenge in SDB, so in this condition PlanetScope with daily temporal resolution capabilities is an optimistic choice. However, this image falls into a relatively new image category. In this study we tested the performance of Sentinel-2A imagery and new PlanetScope imagery bands. The existence of the new sensor owned by PlanetScope allows an increased choice of SDB data sources with high spatial and temporal resolution that is better than the general datasets currently available. There are four additional channels are Coastal Blue, Green I, Yellow, and Red Edge. Assessment is needed to test the capabilities of each new channel using empirical methods via the Stumpf algorithm. Based on accuracy assessment observations, the Sentinel-2A channel combination is still better in terms of accuracy and determination, because it is able to represent the depth in the study area up to 70%. The potential use of PlanetScope's new channels for SDB applications can still be seen in the combination of the Coastal Blue and Yellow channels. This channel combination is still able to represent 47% of the depth variations in the study area.
... RF algorithm is an ensemble learning approach composed of a substantial quantity of independent decision trees (Mudiyanselage et al., 2022). In this algorithm, when the regression model is trained, multiple decision trees are constructed and the average value of all decision trees is output as the final value of the model prediction. ...
Detecting earthquake-induced bathymetric changes helps to understand the geomorphologic process of tufa lakes. Traditional field measurement methods are difficult for spatially complete and continuous bathymetric mapping. Multi-temporal high-resolution optical satellite images are cost-efficient data used for bathymetric change detection. However, for detecting bathymetric changes in tufa lakes, collecting high-density depth calibration data and constructing highly robust water depth inversion models pose certain challenges. This study takes Huohua Lake before and after the Jiuzhaigou Earthquake as the research object, and carries out the bathymetric change detection based on high-resolution remote sensing data. Initially, the WorldView-2 (WV-2) multispectral images obtained before and after the earthquake under the water-storage state of the lake were used as the data source, and the unmanned aerial vehicle (UAV)-based measurement under the water-free state of the lake after the earthquake was used as the bathymetric calibration and validation data. Then using satellite-derived image reflectance, we constructed two-phase bathymetric models with machine learning methods, namely random forest (RF), support vector machine (SVM), and multi-layer perceptron (MLP). The comparison results with classical regression models indicate that machine learning-based water depth inversion models are generally superior. Specifically, the R2 (coefficient of determination) of the optimal model RF reach 0.85 and 0.91, with RMSE (root mean square error) of 1.40 m and 1.08 m. The bathymetric difference maps generated from water depth inversion results reveal that during the period from October 2016 to January 2022, the core area of Huohua Lake experienced more erosion than accretion due to the earthquake-induced flooding. The spatial patterns of changes show that the erosion mainly located in the raised tufa mound area, while the accretion was concentrated in the shallow flat area. This study provides a remote sensing approach for quantifying bathymetric changes in tufa lakes after extreme geological disasters.
