Sea Surface Reflection Research Articles

Estimation of sea surface salinity from MODIS-Aqua data in the northern South China Sea

Sea Surface Salinity (SSS) is a key variable in physical oceanography and biogeochemical research. Microwave remote sensing technology has been used to monitor SSS in open ocean areas. However, the limited spatial resolution and suboptimal precision exhibited by microwave remote sensing derived SSS data have led to the omission of crucial oceanic information, consequently constraining its utilization. In this study, an artificial neural network (ANN) model for the regional SSS inversion was constructed utilizing in situ SSS measurements in the northern South China Sea and remote sensing sea surface temperature and reflectance at wavelengths of 412, 443, 488, 555 and 667 nm derived from MODIS-Aqua satellite Level 2 data. The model showed a good performance of root mean square error (RMSE) of 0.32 and 0.34 for the model training and testing datasets, with mean bias (MB) of 0 and 0.01, and correlation coefficients (R2) reaching 0.93. The model validation indicated that the ANN model reproduced the spatial distribution of SSS highly consistent with the measured SSS (RMSE = 2.75, MB = −1.67, R2 = 0.95), demonstrating its strong applicability in the NSCS. In contrast to other local SSS inversion models, our ANN model clearly showed the fine-scale SSS feature under the intrusion of Pearl River plume near to Dongsha Island with the help of anticyclonic-cyclonic eddies. This prediction closely aligned with the observations from the Soil Moisture Active Passive SSS dataset. The ANN model was applied to MODIS-Aqua data to generate the monthly SSS from 2003 to 2022. A decreasing SSS trend in SSS of 0.02 psu y−1 during 2005–2013 and 2017–2020 and a slightly increasing SSS trend of 0.01 psu y−1 from 2013 to 2017 were observed. Further analysis shows the low-frequency variability of SSS is significantly correlated with the Pacific Decadal Oscillation.

Effects of the Earth curvature on Mie-scattering radiances at high solar-sensor geometries based on Monte Carlo simulations.

Given the importance of vector radiative transfer models in ocean color remote sensing and a lack of suitable models capable of analyzing the Earth curvature effects on Mie-scattering radiances, this study presents an enhanced vector radiative transfer model for a spherical shell atmosphere geometry by the Monte Carlo method (MC-SRTM), considering the effects of Earth curvature, different atmospheric conditions, flat sea surface reflectance, polarization, high solar and sensor geometries, altitudes and wavelengths. A Monte Carlo photon transport model was employed to simulate the vector radiative transfer processes and their effects on the top-of-atmosphere (TOA) radiances. The accuracy of the MC-SRTM was verified by comparing its scalar model outputs from Henyey-Greenstein (HG) phase function with the Kattawar-Adams model results, and the mean relative differences were less than 2.75% and 4.33% for asymmetry factors (g-values) of 0.5 and 0.7, respectively. The vector mode results of MC-SRTM for a spherical shell geometry with the Mie-scattering phase matrix were compared with the PCOART-SA model results (from Polarized Coupled Ocean-Atmosphere Radiative Transfer model based on the pseudo-spherical assumption), and the mean relative differences were less than 2.67% when solar zenith angles (SZAs) > 70 ∘ and sensor viewing zenith angles (VZAs) < 60 ∘ for two aerosol models (coastal and tropospheric models). Based on the MC-SRTM, the effects of Earth curvature on TOA radiances at high SZAs and VZAs were analyzed. For pure aerosol atmosphere, the effects of Earth curvature on TOA radiances reached up to 5.36% for SZAs > 70 ∘ and VZAs < 60 ∘ and reduced to less than 2.60% for SZAs < 70 ∘ and VZAs > 60 ∘. The maximum Earth curvature effect of pure aerosol atmosphere was nearly same (10.06%) as that of the ideal molecule atmosphere. The results also showed no statistically significant differences for the aerosol-molecule mixed and pure aerosol atmospheres. Our study demonstrates that there is a need to consider the Earth curvature effects in the atmospheric correction of satellite ocean color data at high solar and sensor geometries.

Apparent surface-to-sky radiance ratio of natural waters including polarization and aerosol effects: implications for above-water radiometry

Above-water radiometry (AWR) methods have been developed to provide “ground-truth” (or fiducial) measurements for calibration and validation of the water color satellite missions. AWR is also an important tool for environmental survey from dedicated field missions. Under clear sky, the critical step of AWR is to retrieve the water-leaving radiance from radiometric measurements of the upward radiance that also includes the reflection of the direct sunlight and diffuse skylight reflected by the wind ruffled water surface toward the sensor. In order to correct for the surface reflection, sky radiance measurements are performed and converted into surface radiance through a factor often called “sea surface reflectance factor” or “effective Fresnel reflectance coefficient”. Based on theoretical and practical considerations, this factor was renamed surface-to-sky radiance ratio, Rss, to avoid misuse of the term reflectance as often encountered in the literature. Vector radiative transfer computations were performed over the spectral range 350–1,000 nm to provide angular values of Rss for a comprehensive set of aerosol loads and types (including maritime, continental desert and polluted models) and water surface roughness expressed in wave slope variances or in equivalent Cox-Munk wind speeds, for practical use. After separating direct and diffuse light components, it was shown that the spectral shape and amplitude of Rss are very sensitive to aerosol load and type even for extremely low values of the aerosol optical thickness. Uncertainty attached to Rss was computed based on propagation of errors made in aerosol and surface roughness parameters demonstrating the need to adapt the viewing geometry according to the Sun elevation and to associate concurrent aerosol measurements for optimal AWR protocols.

Prediction of Sea Surface Reflectivity under Different Sea Conditions Based on the Clustering of Marine Environmental Parameters

The high-precision prediction of sea clutter reflectivity is helpful in improving the performance of marine radar and sea surface remote sensing capabilities. Under the same sea state, when the significant wave height, wave period, wind speed, and other marine environmental parameters are different, the backward reflectivity of the sea clutter corresponding to the wave structure is not the same. Due to the complex and variable nature of sea clutter characteristics across various wave structures, a meticulous classification of wave structures by integrating multiple marine environmental parameters enables the achievement of the high-precision prediction of sea clutter reflectivity. In this study, utilizing measured data of diverse marine environmental parameters in the Yellow Sea, China, we applied the Affinity Propagation algorithm to data clustering. Based on the clustering outcomes, we accomplished a refined classification of wave structures and developed a discriminant model to precisely classify the refined wave structure, facilitating the categorization of new data. In order to achieve more accurate predictions of sea clutter reflectivity, this paper proposes a deep neural network model named GIT-HYB-DNN, which combines the empirical models GIT and HYB. The GIT-HYB-DNN model is applied to predict the reflectivity for each wave structure category separately. The results demonstrate that the root mean square errors of sea clutter reflectivity predictions for different wave structure categories in this study range from 0.62 dB to 0.84 dB. The prediction errors are significantly reduced compared to the root mean square error of 1.08 dB, which was obtained without refined wave structure classification. This study holds theoretical significance and practical value for the investigation of sea clutter characteristics and the selection of radar parameters.

Retrieval of an On-Orbit Bidirectional Reflectivity Reference in the Mid-Infrared Bands of FY-3D/MERSI-2 Channels 20

The acquisition of high-accuracy reflectance in mid-infrared channels is of great significance for the on-orbit cross-calibration of other bands using the mid-infrared band. However, due to the phenomenon that some sensors have a wide range of wavelengths covered by adjacent channels in the mid-infrared band, the traditional method of estimating the mid-infrared reflectivity assumes that the sea surface reflectivity in different mid-infrared bands is equal, which will lead to a large error during calculation. To solve this problem, this study proposes a nonlinear split-window algorithm involving ocean sun glint data to retrieve reflectivity of FY-3D/MERSI-2 channels 20. The results show that the variation range of sea surface reflectivity of channel 20 in the glint area is 10~25%, the mean value of the reflectivity difference obtained by the nonlinear split-window algorithm is 0.27%, and the RMSE is 0.0066. Among the main influencing factors, the atmospheric conditions have the greatest impact, and the effects of the uncertainties in the water vapor content and aerosol optical thickness on the calculation results are 1.16% and 0.34%, respectively. The initial value limits of the mid-infrared sea surface reflectivity also contribute approximately 0.84%, and their contribution to the uncertainty represents one of the main components. This work shows that the nonlinear split-window algorithm can calculate the infrared sea surface reflectivity with high accuracy and can be used as a reference for in-orbit cross-calibration between different bands.

Impact of Atmospheric Correction on Classification and Quantification of Seagrass Density from WorldView-2 Imagery.

Mapping the seagrass distribution and density in the underwater landscape can improve global Blue Carbon estimates. However, atmospheric absorption and scattering introduce errors in space-based sensors' retrieval of sea surface reflectance, affecting seagrass presence, density, and above-ground carbon () estimates. This study assessed atmospheric correction's impact on mapping seagrass using WorldView-2 satellite imagery from Saint Joseph Bay, Saint George Sound, and Keaton Beach in Florida, USA. Coincident in situ measurements of water-leaving radiance (), optical properties, and seagrass leaf area index (LAI) were collected. Seagrass classification and the retrieval of LAI were compared after empirical line height (ELH) and dark-object subtraction (DOS) methods were used for atmospheric correction. DOS left residual brightness in the blue and green bands but had minimal impact on the seagrass classification accuracy. However, the brighter reflectance values reduced LAI retrievals by up to 50% compared to ELH-corrected images and ground-based observations. This study offers a potential correction for LAI underestimation due to incomplete atmospheric correction, enhancing the retrieval of seagrass density and above-ground Blue Carbon from WorldView-2 imagery without in situ observations for accurate atmospheric interference correction.

Underwater Wireless Sensor Network-Based Localization Method under Mixed Line-of-Sight/Non-Line-of-Sight Conditions

Source localization in underwater sensor networks (UWSNs) presents complex challenges due to sensor nodes drift caused by ocean currents, non-line-of-sight (NLOS) propagation resulting from underwater multipath effects, and environmental noise. This paper proposes a practical and innovative algebraic solution based on the time difference of arrival (TDOA) for source localization in shallow seas. The proposed solution effectively addresses the issues arising from sensor position errors and multipath effects by incorporating the sea-surface reflection non-line-of-sight (SNLOS) link and optimizing the algorithm, thereby significantly improving positioning accuracy. The core concept of the method involves utilizing the weighted least squares algorithm to obtain an initial estimate of the source position, followed by direct estimation of the bias and subsequent refinement of the solution. In contrast to traditional closed-form solutions, this method avoids the introduction of intermediate parameters and directly handles the estimated bias from the previous step. Even when only considering the line-of-sight (LOS) link, the proposed solution achieves precise localization with a minimal number of sensors. Theoretical analysis demonstrates that the solution can achieve the Cramér–Rao lower bound (CRLB) accuracy under low noise conditions, and simulation results validate the superior performance of the proposed solution.

Characteristics of the far-field source signature and its ghost effect of a marine single air gun

The estimation of the air gun source signature is a crucial topic in marine seismic. Typical approaches such as numerical modelling or near-field measurement-based inversion normally work in practice, but suffer from lack of knowledge on the varying sea surface reflection coefficient, the signature directional consistency and the dynamically changed source geometry. For better understanding of the above uncertainties, real far-field air gun source signatures were acquired. A customized approach combined with sparse f-k transformation and adaptive subtraction is applied to deblend the source signature from the sea-floor reflections. Characteristic of the ghost notch distribution and bubble oscillation period indicate an overall good match to the simulated results, except the latter part due to the unexpected air-leakage. Amplitude decay with incidence-angle increase is observed, indicating that the point-source spherical theory for directional signature estimation may be questionable. The sea-surface reflection coefficient appears time-variant and frequency-dependent which challenges the assumption of flat sea surface in signature estimation. An innovative notional source inversion workflow is adapted to the frequency-dependent reflection coefficient. The notional signature inversed through the adapted approach results in less ghost residuals and weaker spectral notch effect comparing with the normal inversion result, i.e., based on a constant reflection coefficient.

Topography-adjusted Monte Carlo simulation of the adjacency effect in remote sensing of coastal and inland waters

We present a Monte Carlo radiative transfer (RT) code that simulates the top-of-atmosphere (TOA) reflectance of waterbodies considering the adjacency effect. The code is the first open-source tool that supports modeling of the adjacency effect with arbitrary topography. It uses the same atmospheric and aerosol settings as 6S, along with user-input surface reflective properties and topography, allowing users to transition from 1D to 3D RT modeling and characterize the adjacency effect in their study areas. The calculation of radiometric quantities was validated against libRadtran, showing a maximum difference lower than 0.6% in extreme optical settings. Examples of the use of the code are presented in three case studies where modeled and measured radiative properties align with each other. One case study shows that 83.7% of the variance in the near-infrared TOA reflectance of 47 lakes in Minnesota was explained by the adjacency effect, emphasizing the significance of the adjacency effect to atmospheric correction algorithms that use near-infrared bands to retrieve aerosol and glint information. Another case study supports the finding of strong wavelength dependence of the effective sea-surface reflectance in above-water measurements of remote sensing reflectance. The code will support physics-based methods that remove the adjacency effect in atmospheric correction processes, and it has the potential to improve satellite-based monitoring of coastal and inland waterbodies.

Simulations of sea surface reflection for V-band O2 differential absorption radar barometry

This study simulates V-band sea surface reflectance and normalized radar cross-section (NRCS) for sea surface air pressure barometry using a differential absorption radar operating at three spectrally even spaced close frequency bands (65.5, 67.75 and 70.0 GHz) with ± 15° cross-track scanning angle. The reflectance ratios of two neighboring frequency pairs and the ratio of the two ratios or three-channel approach are the focus of this study. Impacts of major sea surface geophysical variables such as sea surface temperature, wind, salinity, whitecap, and incidence angle on these reflection properties are analyzed. The reflection simulation is essentially based on geometric optics of rough sea surface. Simulation shows that NRCS values are sufficiently strong within the scanning angle and sea surface salinity would only introduce minimal variations in the surface reflection. The impact of sea surface reflection variations with sea surface temperature, wind, and whitecaps on sea surface barometry are mitigated when the ratios of frequency-paired radar signals are used. Furthermore, the ratios of a three-channel approach are very close to unity and calibration or compensation for the reflectance ratios may not be needed for sea level pressure retrievals. These results improve our understanding of sea surface reflection variations and would help the system design and development.

Calibration of Sentinel-2 Surface Reflectance for Water Quality Modelling in Binh Dinh’s Coastal Zone of Vietnam

Coastal zones are critically important ecosystems that are closely tied to human activities, such as tourism, urbanization, transport, and aquaculture. However, managing and monitoring sea water in the coastal areas is often challenging due to the diversity of the pollution sources. Traditional approaches of onsite measurement and surveys have limitations in terms of cost, efficiency and productivity compared with modern remote sensing methods, particularly for larger and longer observations. Optical remote sensing imagery has been proven to be a good data source for water quality assessment in general and for seawater studies in particular with the use of advanced techniques of data processing such as machine learning (ML) algorithms. However, optical remote sensing data also have their own disadvantages as they are much affected by climatic conditions, atmospheric gas and particles as a source of noise in the data. This noise could be reduced, but it is still unavoidable. This study aims to model seawater quality parameters (total suspended solids (TSS), chlorophyll-a (chla), chemical oxygen demand (COD), and dissolved oxygen (DO)) along a 134 km sea coastal area of the Binh Dinh province by applying the current robust machine learning models of decision tree (DT), random forest (RF), gradient boosting regression (GBR), and Ada boost regression (ABR) using Sentinel-2 imagery. To reduce the atmospheric effects, we conducted onsite measurements of sea surface reflectance (SSR) using the German RAMSES-TriOS instrument for calibration of the Sentinel-2 level 2A data before inputting them to the ML models. Our modeling results showed an improvement of the model accuracy using calibrated SSR compared with the original Sentinel-2 level 2A SSR data. The RF predicted the most accurate seawater quality parameters compared with in situ field-measured data (mean R2 = 0.59 using original Sentinel-2 level 2A SSR and R2 = 0.70 using calibrated SSR). The chla was the most precise estimate (R2 = 0.74 when modelled by the RF model) flowing by DO, COD and TSS. In terms of seawater quality estimation, this accuracy is at a good level. The results of the seawater quality distributions were strongly correlated with coastal features where higher values of TSS, chla, COD, and DO are near the river mouths and urban and tourist areas. These spatial water quality data could be extremely helpful for local governments to make decisions when the modelling is continuously conducted (using big data processing), and it is highly recommended for more applications.

Three-dimensional imaging of ships in the foggy environment using a single-photon detector array

When using single-photon detector array lidar to detect ships in a foggy environment, the multiple scattering between laser and sea fog, sea surface and ships leads to a large number of fake images in the reconstructed depth images, which seriously limits the ability to detect and recognize ships. Here, we propose a three-dimensional imaging algorithm suitable for ships in fog with a single-photon detector array lidar. This algorithm can eliminate fake images caused by multiple reflections of the sea surface in a foggy environment. Firstly, the algorithm uses a dual-Gamma estimation algorithm to eliminate the influence of non-signal photons, and uses a pulse-like selection algorithm to accurately extract the target echo waveform. Secondly, based on the characteristics of the target echo, the principal component analysis (PCA) and K-means algorithm are used to cluster the features of the separated echo signals. Finally, the result of feature classification is used to achieve efficient fake images removal. We show that the proposed algorithm can more accurately reconstruct the depth image of a ship with a distance of 1.72 km in fog environment. Compared with the traditional algorithms, the proposed algorithm can suppress the sea surface reflection while retaining the most target contour. It has been successfully demonstrated for small ships at different distances. This research is expected to improve the detection and recognition ability of single-photon detector array lidar for small targets in complex scenes.

Automating the Acoustic Detection and Characterization of Sea Ice and Surface Waves

Monitoring the status of Arctic marine ecosystems is aided by multi-sensor oceanographic moorings that autonomously collect data year-round. In the northeast Chukchi Sea, an ASL Environmental Sciences Acoustic Zooplankton Fish Profiler (AZFP) collected data from the upper 30 m of the water column every 10–20 s from 2014 to 2020. We here describe the processing of the AZFP’s 455 kHz acoustic backscatter return signal for the purpose of developing methods to assist in characterizing local sea ice conditions. By applying a self-organizing map (SOM) machine learning algorithm to 15-min ensembles of these data, we are able to accurately differentiate between the presence of open water and sea ice, and thereby characterize statistical properties surface wave height envelopes and ice draft. The ability to algorithmically identify small-scale features within the information-dense acoustic dataset enables efficient and rich characterizations of environmental conditions, such as frequency of sparse ice floes in otherwise open water and brief open-water leads amidst the ice pack. Corrections for instrument tilt, speed of sound, and water level allow us to resolve the sea surface reflection interface to within approximately 0.06 ± 0.09 m. By automating the acoustic data processing and alleviating labor- and time-intensive analyses, we extract additional information from the AZFP backscatter data, which is otherwise used for assessing fish and zooplankton densities and behaviors. Beyond applications to new datasets, the approach opens possibilities for the efficient extraction of new information from existing upward-looking sonar records that have been collected in recent decades.

Towards reinforcement learning in UAV relay for anti-jamming maritime communications

Maritime communications with sea surface reflections and sea wave occlusions are susceptible to jamming attacks due to the wide geographical area and intensive wireless communication services. Unmanned Aerial Vehicles (UAVs) help relay messages to improve communication performance, but the relay policy that depends on the rapidly changing maritime environments is difficult to optimize. In this paper, a reinforcement learning-based UAV relay policy for maritime communications is proposed to resist jamming attacks. Based on previous transmission performance, the relay location, the received power of the transmitted signal and the received jamming power, this scheme optimizes the UAV trajectory and relay power to save the energy consumption and decrease the Bit-Error-Rate (BER) of the maritime signals. A deep reinforcement learning-based scheme is also proposed, which designs a deep neural network with dueling architecture to further improve the communication performance and computational complexity. The performance bounds regarding the signal to interference plus noise ratio, energy consumption and the communication utility are provided based on the Nash equilibrium of the game against jamming, and the computational complexity of the proposed schemes is analyzed. Simulation results show that the proposed schemes improve the energy efficiency and decrease the BER compared with the benchmark.

Source and receiver deghosting by demigration‐based supervised learning

AbstractDeghosting of marine seismic data is an important and challenging step in the seismic processing flow. We describe a novel approach to train a supervised convolutional neural network to perform joint source and receiver deghosting of single‐component (hydrophone) data. The training dataset is generated by demigration of stacked depth migrated images into shot gathers with and without ghosts using the actual source and receiver locations from a real survey. To create demigrated data with ghosts, we need an estimate of the depth of the sources and receivers and the reflectivity of the sea surface. In the training process, we systematically perturbed these parameters to create variability in the ghost timing and amplitude and show that this makes the convolutional neural network more robust to variability in source/receiver depth, swells and sea surface reflectivity. We tested the new method on the Marmousi synthetic data and real North Sea field data and show that, in some respects, it performs better than a standard deterministic deghosting method based on least‐squares inversion in the τ‐p domain. On the synthetic data, we also demonstrate the robustness of the new method to variations in swells and sea‐surface reflectivity.

Performance analysis of empirical models of sea surface reflectivity based on measured data

The empirical models of sea surface reflectivity have many categories and different applicable conditions. In order to solve the problem of how to select the best models in different environments, we make a systematic analysis of the performance of different models based on the measured data of sea surface and the meteorological observation data in the Northwest Pacific. The results show that the accuracy of the empirical models of sea surface reflectivity may be biased when only the sea state is used as the description of sea surface roughness. For low radar grazing angle, NRL and HYB models can obtain better performance. It is recommended to add a fixed offset for NRL model at medium grazing angles. Constant gamma model can get better fit at high radar grazing angle.

An MPI-accelerated Monte Carlo algorithm for estimating the reflectance and transmittance properties of a wind-driven sea surface

An MPI-accelerated Monte Carlo algorithm for estimating the reflectance and transmittance properties of a wind-driven sea surface

Modeling of input signals reflected from coastal zones and observed by small-sized radar systems from UAVs

When developing modern unmanned aerial vehicle-based onboard small-sized radar systems, which are used for monitoring of coastal zones for their operational and environmental purposes, adequate mathematical models are needed of reflections from the corresponding resolution elements of the underlying surfaces, located on the water–land interface. We consider the development of some mathematical models for such applications. We analyze two modern approaches used in developing mathematical models for input signals reflected both separately from the terrestrial and marine surfaces and simultaneously from two aforementioned surfaces at once. A statistical approach and two families of distributions are chosen for modeling the signals, which enable synthesis of efficient algorithms for modeling the fluxes of random variables. The input signals of sea surface reflections are approximated by the log-normal law, while the reflections from the land surface are approximated by the Weibull distribution. For edge-coastal modeling, it is proposed to use the law of distribution of the vector sum of the flux of random variables comprising the input signals reflected from elementary land and sea areas that simultaneously fall within the corresponding resolution element. A highlight of the models developed is the consideration of both the probability distribution laws of the reflected signals and their correlation-spectral characteristics between resolution elements, as well as the anisotropy of reflections when monitoring surfaces from several angles. The presented mathematical models are consistent with experimental data sets of reflections, obtained from coastal zones. The development of coastal reflection models and corresponding modeling algorithms reduces the required number of actual field tests and saves time for implementation of monitoring systems, and hence, limits the overhead of this implementation.

GRNN-Based Predictors of UHF-Band Sea Clutter Reflectivity at Low Grazing Angle

As a basic characteristic of sea clutter, the reflectivity of sea surface depends on many factors. Various universal empirical models of low precision have been developed to predict the reflectivity of sea surface. In this letter, a method is proposed to train specific predictors by big data learning, where the universal empirical models are embedded to the architecture of the generalized regression neural network (GRNN) to enhance the learning ability and efficiency. On the sea clutter database measured by an island-based UHF-band radar in the offshore waters of the Yellow sea of China at low grazing angle, the GRNN-based predictors of different structures are compared with other predictors. The results on the database show that the GRNN-based predictors behave better at learning efficiency, prediction precision, and robustness.

An APMLP Deep Learning Model for Bathymetry Retrieval Using Adjacent Pixels

Shallowwater depth plays an important role in marine development, navigation safety, and environmental protection. It is an efficient and economical way to obtain water depth by remote sensing technology. At present, most empirical models based on multispectral image usually obtain water depth by the relationship between the sea surface reflectance (SSR) (a single pixel) and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in situ</i> water depth, it is a one-to-one correspondence between the reflectance and depth. However, seafloor substrate and inherent optical properties (IOP) will also have contribution to the SSR. In this article, we propose an adjacent pixels multilayer perceptron model (APMLP) model using adjacent pixels to weaken the influence of seafloor substrate and IOP.Datasets on Oahu Island (Sentinel-2B, LIDAR <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in situ</i> data) and Saint Thomas Island (Sentinel-2A, LIDAR <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in situ</i> data) are used to establish and verify the model. The APMLP model are also compared with the multilayer perceptron model (MLP) model, BP neural network model, and Log-ratio model. The overall root-mean-square error (RMSE) of APMLP model on Oahu Island is 0.72 m, which is much better than the other three models (MLP 1.07 m, BP 1.05 m, Log-ratio 1.52 m). Similar results are obtained from the Saint Thomas Island dataset, RMSE of APMLP model is 1.56 m, better than the other three (MLP 1.91 m, BP 1.89 m, Log-ratio 2.39 m). The study confirms that considering adjacent pixels in an artificial neural network model can effectively improve the performance of water depth retrieval.