Calder Brian’s scientific contributions

Significant amounts of labor-intensive effort and time are needed for compiling and maintaining Electronic Navigation Charts (ENCs). The great amount of data collected with high-resolution systems that are being delivered to the charting divisions along with initiatives within the divisions, such as the Office of Coast Survey (OCS)/ Marine Chart Division's (MCD) rescheme project, often lead to a bottleneck situation. Therefore, nowadays, one of the main objectives in many Hydrographic Offices (HOs) is automating generalization tasks to improve productivity in a cost-effective manner. However, regardless of the various relevant research efforts and the advancements in technology, most generalization tasks are still performed manually or semi-manually, where a likelihood of human error is admitted. The ideal situation would be a fully automated solution which could minimize the time and effort needed for ENC production and support rapid chart updates. Full automation can streamline tasks such as the re-compilation of bathymetry and the migration from the imperial to the metric system for the OCS/MCD rescheme project. In addition, the time a fully automated solution would save could, indirectly, solve other compilation problems with allowing cartographers and HOs to steer their focus towards fixing them, such as the edge matching inconsistencies between adjacent ENCs. Towards this optimum goal, we present a research work that aims to translate cartographic practice and theory into algorithmic building blocks that can iterate and cooperate to find the appropriate chart representation for any given area, at any scale, optimized according to set criteria. Toward the ultimate goal of full automation in chart compilation through scales, previous research efforts were investigated, and available nautical cartographic specifications were reviewed. The thereinto generalization guidelines were extracted, categorized and translated into rules to be defined in a constraints template as conditions to be respected during the generalization process. Accordingly, an automated nautical generalization (ANG) model was developed to form a comprehensive process that utilizes the generated template, as the input that derives the data generalization for any desired output scale, and the largest scale chart data to perform the generalization to the target scale respecting topology constraints. Lastly, since safety is the ultimate product constraint in the domain, a custom validation tool detects any safety violation for the generalized linear features in the output database for user correction.

Airborne lidar bathymetry (ALB) and multi-beam echo-sounders (MBES) are two commonly used methods employed by the hydrographic community for bathymetric data collection. Although these technologies can collect high resolution and accurate depth data, they present some limitations related to whether, cost, time, and survey difficulties in near-shore environments. Satellite Derived Bathymetry (SDB) is a relatively new promising technique that utilizes optical satellite remote sensing sensors for estimating bathymetry in shallow water areas. The practice in previous works is with considering a linear relationship between calibrations points and the SDB model, assuming null water column and constant bottom type returns, while a single global bathymetric model is calibrated for the entire scene. To address areas where these assumptions fail and non-uniform water column or bottom conditions can affect the performance of conventional global models, this work presents a vertically and horizontally adaptive model to improve bottom depth estimation. With a case study in Dry-Tortugas, Florida and utilizing Sentinel-2A image, lidar, and chart data in depths up to 30m, we demonstrate the ability of a locally calibrated, piecewise linear function, in providing improved bathymetric estimates compared to conventional techniques. The improved accuracy of the bathymetric estimates with the presented geographically adapted model can enhance the use of satellite derived bathymetry for nautical chart purposes. Figure: Workflow for deriving bathymetry using geographically adaptive model.

Current methods for nautical charts generalization are strongly human interactive and time-consuming. Significant amounts of effort are needed for generalizing, compiling, updating, maintaining, and consistency checking those products. The ideal situation would be a fully automated solution for generating products on demand, at the right scale, at the point of use, and directly from the seamless database, that respect all application specific constraints (i.e., Safety, Legibility, Topology, and Morphology). Such a solution would solve many related problems, minimize the time and effort needed for ENC production and support rapid update. However, regardless of the advancements in technology and the various research efforts, generalization tasks for nautical chart compilation are still performed mostly manually or semi-manually, where a likelihood of human error must be admitted. Furthermore, practice has shown that there is often a trade off among the various generalization constraints as fixing one violation may result in the violation of another constraint. Towards this optimum goal, we are conducting a project that aims to investigate the previous efforts for automated map production, review the available relevant nautical cartographic standards and specifications, and extract and categorize the nautical chart generalization guidelines thereinto. These are subsequently translated into rules and defined in a template as conditions to be respected during the generalization process. According to a hierarchy level, a cost will be related to any violation according to its importance to the safety of navigation. Since fully satisfying all constraints seems infeasible, the optimal generalized chart should be the one with the lowest cost. For the implementation, a multi-agent generalization model is under development in the ESRI nautical environment that will utilize the template and the source data within the areas of interest to perform the generalization for the target scale (Figure). Figure: The concept of the nautical generalization model