Hydrographic Data Processing: Cleaning and Sound-Velocity Correction

Hydrographic Data Processing: Cleaning and Sound-Velocity Correction

Why Raw Soundings Need Processing

When a single-beam echosounder records depths during a field survey, it captures everything: valid bottom returns, acoustic noise, fish schools, debris, air bubbles near the transducer, and momentary GPS dropouts. Before that data can be used to produce a bathymetric chart or a design surface, it must be cleaned and corrected so that every depth value represents a true, physical bottom measurement referenced to the correct datum. This lesson walks through the three main corrections—draft, sound velocity, and outlier removal—and explains how to document your edits so the dataset is defensible.

Draft Correction

The transducer is mounted below the waterline on the survey vessel. The vertical distance from the water surface to the face of the transducer is called the vessel draft offset (or transducer draft). Because the echosounder measures range from the transducer face to the bottom, the reported depth is shallower than the true water depth by that offset. You add the draft value to every raw sounding to recover the true depth below the water surface.

For example, if the transducer sits 0.35 m below the waterline and the echosounder records 4.20 m, the corrected depth is 4.20 + 0.35 = 4.55 m. Draft is entered as a constant offset in the processing software during project setup. Verify this value against the field notes before applying it; using the wrong draft affects every sounding in the file.

Sound-Velocity Correction

An echosounder calculates depth by timing the round-trip travel of an acoustic pulse and multiplying by the assumed speed of sound in water, then dividing by two:

> Depth = (travel time × assumed velocity) / 2

The instrument is typically calibrated assuming a single velocity value, often 1500 m/s. Actual sound speed in water depends on temperature, salinity, and pressure (depth). If the true average velocity differs from the assumed value, every depth reading is in error by a proportional amount.

The correction is applied as a ratio:

> Corrected depth = Raw depth × (true velocity / assumed velocity)

The true average velocity comes from a sound-velocity cast (also called a cast file or SVP file), in which a probe is lowered through the water column to measure velocity at multiple depths. In shallow freshwater surveys, a single representative velocity measured at mid-depth is often sufficient. The cast file is imported into the processing software and the software re-calculates all depths using the measured velocity profile.

Importing Raw Data and Verifying the Vessel Track

Processing software such as HYPACK or CARIS Ping-to-Chart reads raw field files (formats such as .hsx, .hsf, or ASCII logs). After import, the first quality check is to display the vessel track and confirm it forms continuous, logical lines over the planned survey area. Look for:

• Position jumps: sudden position shifts caused by GNSS dropout; the track will show a spike or gap.
• Timestamp mismatches: sounding records and GNSS position records should agree within one second. Larger gaps mean depth and position cannot be reliably matched.
• Missing lines: areas that were not covered during field operations.

Flag or delete position records that fail these checks before applying any depth corrections. Processing software maintains an audit log of edits; never delete records outside the software.

Noise Filtering and Spike Editing

Once corrections are applied, the dataset is inspected for outlier soundings. Two methods are used together:

Statistical filter: The software compares each sounding to its neighbors within a defined window and rejects values that deviate beyond a set threshold (for example, more than a specified number of standard deviations or a fixed distance in meters). This catches systematic noise.

Manual spike editor: The operator views the sounding profile as a cross-section and visually identifies spikes—individual points that are clearly above or below the trend. Common causes include a fish return (shallower than bottom), a sub-bottom reflection (deeper than bottom), or a bubble event near the transducer.

Every flagged sounding must receive a reason code so the edit history is transparent. Typical codes include NOISE, FISH, BUBBLE, POSITION, and OPERATOR. A rejection rate above roughly 5% of total soundings is a signal to investigate further—it may indicate equipment malfunction, poor water conditions, or a calibration error.

Exporting the Cleaned Dataset

The final cleaned dataset is exported in two standard forms:

1. ASCII XYZ file: columns of Easting, Northing, and corrected depth (or elevation). This is universally readable by GIS and CAD software, including QGIS and AutoCAD Civil 3D.

2. Shapefile of the vessel track: a polyline or point file documenting where the vessel traveled, used for coverage verification and report documentation.

A sounding summary table is generated listing total soundings collected, soundings rejected, and the rejection percentage. This table becomes part of the survey report and should be signed by the operator.

Safety and Data Integrity Notes

Hydrographic data products are used to design navigation channels, assess flood risk, and locate utilities under water. An uncorrected or carelessly edited dataset can result in unsafe navigation clearances or incorrect engineering decisions. Treat every edit as professionally binding. Back up raw files before any processing session—work only on copies. If field notes are incomplete (draft offset unrecorded, cast not taken), document the gap and flag the deliverable as provisional until the missing information is confirmed.

When working near water during data collection, always follow vessel safety protocols, wear a Coast Guard–approved personal flotation device (PFD), and comply with any applicable waterway regulations. In the lab, follow your institution's policies for data storage and handling of licensed software.