Abstract:The coastal shallow water zone can be a challenging and costly environment in which to acquire bathymetry and other oceanographic data using traditional survey methods. Much of the coastal shallow water zone worldwide remains unmapped using recent techniques and is, therefore, poorly understood. Optical satellite imagery is proving to be a useful tool in predicting water depth in coastal zones, particularly in conjunction with other standard datasets, though its quality and accuracy remains largely unconstrained. A common challenge in any prediction study is to choose a small but representative group of predictors, one of which can be determined as the best. In this respect, exploratory analyses are used to guide the make-up of this group, where we choose to compare a basic non-spatial model versus four spatial alternatives, each catering for a variety of spatial effects. Using one instance of RapidEye satellite imagery, we show that all four spatial models show better adjustments than the non-spatial model in the water depth predictions, with the best predictor yielding a correlation coefficient of actual versus predicted at 0.985. All five predictors also factor in the influence of bottom type in explaining water depth variation. However, the prediction ranges are too large to be used in high accuracy bathymetry products such as navigation charts; nevertheless, they are considered beneficial in a variety of other OPEN ACCESSRemote Sens. 2015, 7 13783 applications in sensitive disciplines such as environmental monitoring, seabed mapping, or coastal zone management.

Bathymetry estimated from optical satellite imagery has been increasingly implemented as an alternative to traditional bathymetric survey techniques. The availability of new sensors such as Sentinel-2 with improved spatial and temporal resolution, in comparison with previous optical sensors, offers innovative capabilities for bathymetry derivation. This study presents an assessment of the fit between satellite data and the underlying models in the most widely used empirical algorithms: the linear band model and the log-transformed band ratio model using Sentinel-2A data. Both models were tested in two study areas of the Irish coast with different morphological and environmental conditions. Results showed that the linear band model fitted better than the log-transformed band ratio model providing coefficient of determination values, R 2 , between 0.83 and 0.88 (0 m-10 m) for the five images considered in the study. The closest fit was found in the depth range 2 m-6 m. Atmospheric correction, bottom type influence, and water column conditions proved to be key factors in the bathymetric derivation using these satellite datasets.

Optical satellite data is an efficient and complementary method to hydrographic surveys for deriving bathymetry in shallow coastal waters. Empirical approaches (in particular, the models of Stumpf and Lyzenga) provide a practical methodology to derive bathymetric information from remote sensing. Recent studies, however, have focused on enhancing the performance of such empirical approaches by extending them via spatial information. In this study, the relationship between multibeam depth and Sentinel-2 image bands was analyzed in an optically complex environment using the spatial predictor of kriging with an external drift (KED), where its external drift component was estimated: a) by a ratio of log-transformed bands based on Stumpf's model (KED_S) and b) by a log-linear transform based on Lyzenga's model (KED_L). Through the calibration of KED models, the study objectives were: 1) to better understand the empirical relationship between Sentinel-2 multispectral satellite reflectance and depth, 2) to test the robustness of KED to derive bathymetry in a multitemporal series of Sentinel-2 images and multibeam data, and 3) to compare the performance of KED against the existing non-spatial models described by Stumpf et al. and Lyzenga. Results showed that KED could improve prediction accuracy with a decrease in RMSE of 89% and 88%, and an increase in R 2 of 27% and 14%, over the Stumpf and Lyzenga models, respectively. The decrease in RMSE provides a worthwhile improvement in accuracy, where results showed effective prediction of depth up to 6 m. However, the presence of higher concentrations of suspended materials, especially river plumes, can reduce this threshold to 4 m. As would be expected, prediction accuracy could be improved through the removal of outliers, which were mainly located in the channel of the river, areas influenced by the river plume, abrupt topography, but also very shallow areas close to the shoreline. These areas have been identified as conflictive zones where satellite-derived bathymetry can be compromised.

The ability to track the dynamics of processes in natural water bodies on a global scale, and at a resolution that enables highly localised behaviour to be visualized, is an ideal scenario for understanding how local events can influence the global environment. While advances in in-situ chem/bio-sensing continue to be reported, costs and reliability issues still inhibit the implementation of large-scale deployments. In contrast, physical parameters like surface temperature can be tracked on a global scale using satellite remote sensing, and locally at high resolution via flyovers and drones using multi-spectral imaging. In this study, we show how a much more complete picture of submarine and intertidal groundwater discharge patterns in Kinvara Bay, Galway can be achieved using a fusion of data collected from the Earth Observation satellite (Landsat 8), small aircraft and in-situ sensors. Over the course of the four-day field campaign, over 65,000 in-situ temperatures, salinity and nutrient measurements were collected in parallel with high-resolution thermal imaging from aircraft flyovers. The processed in-situ data show highly correlated patterns between temperature and salinity at the southern end of the bay where freshwater springs can be identified at low tide. Salinity values range from 1 to 2 ppt at the southern end of the bay to 30 ppt at the mouth of the bay, indicating the presence of a freshwater wedge. The data clearly show that temperature differences can be used to track the dynamics of freshwater and seawater mixing in the inner bay region. This outcome suggests that combining the tremendous spatial density and wide geographical reach of remote temperature sensing (using drones, flyovers and satellites) with ground-truthing via appropriately located in-situ sensors (temperature, salinity, chemical, and biological) can produce a much more complete and accurate picture of the water dynamics than each modality used in isolation.

Satellite derived bathymetry (SDB) enables rapid mapping of large coastal areas through measurement of optical penetration of the water column. The resolution of bathymetric mapping and achievable horizontal and vertical accuracies vary but generally, all SDB outputs are constrained by sensor type, water quality and other environmental conditions. Efforts to improve accuracy include physics-based methods (similar to radiative transfer models e.g. for atmospheric/vegetation studies) or detailed in-situ sampling of the seabed and water column, but the spatial component of SDB measurements is often under-utilised in SDB workflows despite promising results suggesting potential to improve accuracy significantly. In this study, a selection of satellite datasets (Landsat 8, RapidEye and Pleiades) at different spatial and spectral resolutions were tested using a log ratio transform to derive bathymetry in an Atlantic coastal embayment. A series of non-spatial and spatial linear analyses were then conducted and their influence on SDB prediction accuracy was assessed in addition to the significance of each model's parameters. Landsat 8 (30 m pixel size) performed relatively weak with the non-spatial model, but showed the best results with the spatial model. However, the highest spatial resolution imagery used – Pleiades (2 m pixel size) showed good results across both non-spatial and spatial models which suggests a suitability for SDB prediction at a higher spatial resolution than the others. In all cases, the spatial models were able to constrain the prediction differences at increased water depths.