Automated Identification of Discrepancies between Nautical Charts and Survey Soundings

Timely and accurate identification of change detection for areas depicted on nautical charts constitutes a key task for marine cartographic agencies in supporting maritime safety. Such a task is usually achieved through manual or semi-automated processes, based on best practices developed over the years requiring a substantial level of human commitment (i.e., to visually compare the chart with the new collected data or to analyze the result of intermediate products). This work describes an algorithm that aims to largely automate the change identification process as well as to reduce its subjective component. Through the selective derivation of a set of depth points from a nautical chart, a triangulated irregular network is created to apply a preliminary tilted-triangle test to all the input survey soundings. Given the complexity of a modern nautical chart, a set of feature-specific, point-in-polygon tests are then performed. As output, the algorithm provides danger-to-navigation candidates, chart discrepancies, and a subset of features that requires human evaluation. The algorithm has been successfully tested with real-world electronic navigational charts and survey datasets. In parallel to the research development, a prototype application implementing the algorithm was created and made publicly available.

corresponding paper chart but can embed supplementary material present in other kinds of nautical 116 publications (e.g., sailing directions). The ENC content may be directly derived from original survey 117 material, but also from databased information, existing paper charts or a combination of them [30].

118
The International Hydrographic Organization (IHO) has created a standard vector format, 119 named S-57, for official ENCs (that is, produced by a government hydrographic office) that contain a 120 set of data layers for a range of hydrographic applications. However, an IHO S-57 ENC is mainly 121 used, in combination with positional information from navigation sensors and radar data, by an 122 Electronic Chart Display and Information System (ECDIS) to provide a graphical representation of a 123 marine area, including bathymetry and topography, and to assist the mariner in route planning and 124 monitoring by providing natural features, man-made structures, coastlines, and any other 125 information related to marine navigation [6,31]. To meet high standards of reliability and 126 performance, an extensive body of rules defines an ECDIS with the result that, under SOLAS Chapter 127 V regulations, an ECDIS loaded with official ENCs is currently the only alternative for the navigator 128 to the adequate and up-to-date paper charts [32].

129
As defined by the IHO standard, the S-57 data content consists of a set of features that may have 130 a spatial representation in the form of points, lines or polygons. Based on the ENC product 131 specifications, those features must be encoded using the chain-node topology ( Figure 1) [31,33]. For 132 a S-57 ENC, the basic unit of geographic coverage is called "cell", normally a spherical rectangle 133 bordered by meridians and latitudes with actual data coverage that can be of any shape [30]. Within 134 the same navigational purpose (e.g., coastal, approach) (Table 1), the cell data may not overlap with those of the adjacent cells. The navigational purpose also drives the compilation scale and, thus, the

156
The algorithm described in this work was developed to handle as input the following types of  "Pure survey soundings", intended as the point cloud measured by a bathymetric sonar after 159 having been properly integrated with ancillary sensors, cleaned by spurious measurements and 160 reduced to the chart datum.
 "Selected gridded survey soundings", commonly generated during the cartographic processing 164 using algorithms that select the grid nodes that are more meaningful to depict the trend of the 165 underlining bathymetric model.

166
Assuming the correctness in the generation of the above three types of survey soundings (whose

188
For the bathymetric model reconstruction, the algorithm retrieves nodes from the following 189 ENC feature types (in parenthesis, the adopted S-57 acronym):   (1) For example, with a value of 1:10,000, the obtained by applying (1) is 100 meters.

202
The algorithm retrieves the for (1) directly from the ENC, using the value stored in the

230
Once the TIN has been generated, the input survey soundings are categorized after the 231 identification of the containing triangle. If a specific survey sounding is outside of all the TIN triangles the tilted plane, ( , , ), containing the three triangle vertices, , , : where: The intersection point, , between the vertical line, , containing the point: and the identified plane is calculated using [41]: where is the normal to the tilted plane: The vertical distance, Δ , is calculated as the difference between the depth dimensions 242 between and : The algorithm uses the Δ to evaluate whether flagging the soundings as DtoN candidates 244 ( Figure 4) or potential chart discrepancies. The default logic of the applied thresholding (Table 2)    Although large negative Δ values ("deeps") do not represent a concern from a safety of 264 navigation perspective, their presence provides evidence of chart inconsistencies. As such, the 265 algorithm can be optionally set to identify those deeps, by simply applying a reversed thresholding 266 logic in respect to the "shallow" chart discrepancies.

288
(that is, with three vertices having a 10-meter depth) shown in magenta in the pane (b). By adopting 289 the sounding-in-specific-feature test, the algorithm flagged the survey sounding as a potential 290 discrepancy (shown as grey circle with a depth difference of 1.7 meters in red) since it was detected 291 as contained by a depth area with a valid depth range between ten and twenty meters.

292
The S-57 ENC specifications prescribe that a cell must have the entire area with data covered, 293 without any overlap, by a restricted group of seven area feature types, called the "skin of the earth"

306
The library has been coupled with a multi-tab graphical user interface based on a Python port

323
During the initial development, ad-hoc input survey soundings have been generated with two

324
DtoNs and two chart discrepancies to be tested against a valid ENC. Using the described default 325 settings, the algorithm was able to identify both the DtoNs and the chart discrepancies

329
The nodes are represented as dots, with the connecting triangle edges in gray. Axes in geographical 330 WGS84 coordinates.

331
The algorithm has then been tested in several real-data scenarios using official ENCs and MBES-332 based survey datasets. To keep this section concise, only two of those scenarios are here presented.

333
On one side, the analysis of the soundings from NOAA survey H12990 (see Figure 8) shows relatively 334 minor changes to the current ENCs. Conversely, the presence of large active glaciers in the area is 335 highlighted by the analysis of the NOAA H13701 dataset. Further details of the adopted inputs are 336 provided, for the surveys, in Table 4 and, for the ENCs, in Table 5.

354
Axes in geographical WGS84 coordinates. 355 356 Figure 11. Algorithm results using the H12990 survey soundings compared to the US5VA22M ENC.

358
The results of the analysis for the H13701 survey dataset is shown in Figure 12.

401
In alternative to the use of the Delaunay conforming triangulation, future works may explore 402 the creation of the TIN based on a constrained Delaunay triangulation [58,59], since this technique 403 makes possible to explicitly define edges ("breaklines") that are not modified by the triangulator 404 (thus removing the currently required addition of interpolated nodes along long edges).

405
The conversion by interpolation of a TIN into a regular grid to perform discrepancy analysis 406 represents a quite popular operation. However, although the several intrinsic advantages related to 407 deal with grid data (e.g., higher processing and development speed) [