Learning center

Bathymetry: coordinate conventions, displayed points and complete arrays

Follow all 82,970 Baja rows through explicit depth signs, display limits, missing reference results and distinct viewer/array exports.

Written guide to one descriptive sounding workflow. Engineering Signal Processing’s Bathymetry app can display imported soundings and summarize their input contract. This Baja example follows 82,970 rows through coordinate and depth conventions, bounded plots and a complete private array export. Its two selected routes do not establish survey accuracy or certification.

1. Keep coordinates, depth signs and preparation explicit

The source is the GMT tutorial’s Baja California bathymetry distributed in Verde. The official dataset documentation identifies longitude, latitude and bathymetry in metres. The saved example pins the v1.9.0 compressed source file by hash; the documentation’s “latest” URL does not replace that input identity.

The prepared baja-full-wgs84.csv has three columns: longitude_deg, latitude_deg and elevation_m. It retains all 82,970 rows and 248,910 values in source order. Preparation explicitly subtracts 360 from the source’s 0–360-degree longitude convention to obtain −180–180 degrees. Latitude and negative elevation text are copied unchanged. There is no resampling or invented timing metadata.

This source uses negative elevation below the reference surface. Select the positive-up elevation input convention so the app’s analysis copy contains positive-down depths. That sign conversion is not a tide, heave, draft or vertical-datum correction. Keep the original elevation file and the declared convention with the results.

The saved preparation leaves dataset redistribution terms under review. This guide links to the upstream documentation and records identities; it does not supply a new copy of the dataset for download.

2. Set the contract before importing

  1. Open Bathymetry in Engineering Signal Processing (bathymetry). Set Numeric table schema to xyz_depth.
  2. Set Coordinate contract to longitude_latitude_wgs84. These coordinates are degrees. The default projected-X/Y-in-metres contract would give them a different meaning.
  3. Set Input z/depth convention to positive_up_elevation. Use IMPORT BATHYMETRY to select the prepared CSV. These settings matter during import; confirm the resolved contract in the output.
  4. Select only Seafloor Sounding Map and Bathymetry Input and Provenance QC (routes seafloor_map and summary), then run and open the viewer.

This app-specific import is a row-table workflow. It does not require a waveform sample rate. The recorded run supplied no uncertainty, independent reference depths, acquisition-time column or survey quality labels.

The current geographic contract keeps the native longitude/latitude and creates a local azimuthal-equidistant coordinate copy in metres for applicable calculations. The saved QC reports that analysis CRS. It does not establish the source’s vertical datum, calibration or a new accuracy certificate. This guide checks the saved metric-array identities; it does not rerun or independently validate the projection.

3. Interpret validity and missing reference results

The saved QC reports 82,970 total rows, 82,970 valid rows, zero invalid rows and a valid fraction of 1.0. Its validity source is derived_from_finite_core_xyz: the core coordinates and elevation are finite. This checks the numerical input contract. It does not establish that every sounding is accurate, free of outliers or approved by a hydrographic surveyor.

There is no supplied grid and no optional typed evidence in this example. Both displayed Validated and survey certification fields are false. References to IHO, BAG or NOAA concepts in the result metadata do not turn these descriptive routes into standards certification.

The sounding-map result has reference_depth_rmse_m equal to null in JSON and an empty value in CSV. That means no reference-depth comparison was available; it is not zero error. Likewise, grid_shape is null, and the optional-field list is empty. Preserve these missing-data distinctions when loading exports into a spreadsheet.

4. Distinguish plotted points from retained data

RepresentationScope in this example
Input and summary calculationsAll 82,970 valid source rows. Positive-down depth ranges from 9 to 7,708 m, with a span of 7,699 m.
Sounding-map scatter60,000 selected points. The result reports both valid_depth_rows and display_rows.
Depth along metric X orderUp to 8,000 selected rows after sorting valid rows by metric X. The connected line is not a declared vessel track or an along-track survey profile.
Depth histogramAll valid depths, grouped into 40 bins. Bar heights count soundings; they are not a sea-floor area estimate.
Private array packageSix complete arrays of 82,970 entries each, including native coordinates, analysis depths, validity flags and the metric coordinate copy.

The plotting helper chooses indices across the ordered rows to fit its display limit. It does not generate substitute soundings or establish statistically representative sampling. A plot’s visible density depends on that selection, overplotting and the original sampling pattern. Use the retained arrays when a question requires every sounding.

The earlier independent receipt checked ten source-count and descriptive-depth fields against saved scalar arithmetic. This guide reopens those results and the exports. It does not perform gridding, terrain fitting, uncertainty estimation or a new comparison with independent measured depths.

5. Save the viewer snapshot and the full array package separately

Open Viewer → Export Package saves an inspectable result snapshot. For this example, the ZIP contains two figures, two CSV/JSON table pairs, summary text and a manifest. The tables have 11 sounding-map fields and 25 input/QC fields. The manifest’s complete-table claim applies to those summary tables; it does not mean the ZIP contains an 82,970-row sounding table.

The saved viewer CSV and JSON agree in all 72 cells. The historical native/server comparison agrees in 33 fields, explicitly excluding the two per-run elapsed-time strings and the input-container description. The guide retains those three differences rather than claiming byte-identical native and server runs.

Use the app’s separate EXPORT EVIDENCE PACKAGE action for the full arrays. The reviewed output is one compressed NPZ file containing these six arrays plus a UTF-8 JSON manifest:

ArrayMeaning
x_native
y_native
The prepared longitude and latitude in original row order, in degrees for this contract.
depth_positive_down_mThe analysis depth in metres: the negative of this input’s elevation.
valid_flagOne byte per row; all values are 1 in this example. Interpret using the recorded validity source.
x_metric_analysis_m
y_metric_analysis_m
The saved analysis-coordinate copy in metres. These are distinct from the native longitude/latitude.

Reopening this saved NPZ confirms 331,880 coordinate, depth-sign and flag values against the prepared rows. The other 165,940 metric-coordinate values are complete, finite and match their saved array hashes; no independent reprojection is claimed. The embedded manifest records each array’s type, shape and hash, together with the input contract and run report.

The viewer manifest reports reloadable session: false and does not separately include numeric plot arrays. The private NPZ preserves analysis arrays and declarations, not the original source file, its byte layout or a demonstrated restorable GUI session. Keep the prepared CSV and preparation/source receipt separately. Source paths are withheld from the NPZ, but coordinates and scientific content remain present and still need review before sharing.

Worked examples describe particular inputs, methods and recorded results. Check current availability and input requirements before planning an evaluation.

← Back to the learning center