Day 29: MBES Data Cleaning (Swath Editor) – Masterpiece Edition | River Warrior

🧹 DAY 29: MBES DATA CLEANING (SWATH EDITOR)

⏱️ Estimated Reading Time: 15 Minutes | 🎓 Level: Professional Hydrographer

From Noisy Swaths to Pristine Bathymetry – Outlier Rejection & Quality Control

Instructor: Engr. Rokib Hossain | River Warrior Academy

🏠 Course Homepage

📖 Table of Contents (Serialised)

Why Data Cleaning Is Essential
Types of Outliers in MBES Data
Swath Editor Interface Overview
Manual Cleaning: Point Rejection
Automatic Filters: Slope, Beam Pattern, Statistical
Interactive Cleaning Simulator
Quality Control After Cleaning
Data Cleaning Checklist
Resources & Software Tools
Frequently Asked Questions
Action Items & Next Steps

1. Why Data Cleaning Is Essential

Raw multibeam soundings contain outliers caused by noise, bubbles, fish, acoustic interference, and false bottom detections. If not removed, these outliers distort the final bathymetric surface and can create false features (e.g., artificial peaks or holes). Cleaning (also called swath editing) is the process of identifying and rejecting these erroneous points.

Modern processing software provides both manual and automatic cleaning tools. A good hydrographer knows when to trust automatic filters and when to manually inspect.

🧠 Golden Rule: Clean conservatively – never reject more than 5‑10% of soundings. If you need to reject more, revisit your acquisition parameters (gain, SVP, speed).

🌊 River Warrior Pro-Tip: The Jamuna Current Lesson

During a high‑flow survey in the Jamuna River, strong currents caused cavitation bubbles near the transducer. The side beams produced thousands of false outliers. Instead of manual cleaning, we slowed the vessel from 6 to 3 knots, which eliminated 90% of the noise. Always check physical conditions before intensive cleaning.

2. Types of Outliers in MBES Data

.htmlNear‑nadir cluster那样Group of shallow points below the vessel那样Fish or bubble clouds.htmlOuter beam striping那样Systematic deepening at swath edges那样SVP error, bad roll calibration.htmlAcross‑track line of points那样Flat line crossing the swath那样False bottom detection (sidelobe interference)

Outlier Type	Appearance	Common Cause
Isolated spike那样Single point much shallower or deeper than neighbours那样Propeller noise, electrical spike

Swath profile with typical outliers: isolated spike, SVP edge artifacts, and outer‑beam striping.

3. Swath Editor Interface Overview

The swath editor (available in Qimera, CARIS HIPS, Hypack, MB‑System) displays the soundings in three common views:

Ping view (along‑track): Each vertical line is a ping, showing depth vs beam angle.
Swath view (across‑track): Profile across the swath at a selected ping.
Bathymetric surface view: Colour‑coded depth grid.

Most editors allow you to click on outliers to reject them. Changes are saved in a “flag” file (e.g., .filt or .flags) without altering raw data.

4. Manual Cleaning: Point Rejection

Manual cleaning is time‑consuming but necessary for complex noise. Steps:

Open the swath editor and navigate to suspect areas (use coverage map to find spikes).
Select the ping or beam where the outlier appears.
Click “Reject” or press a hotkey (e.g., ‘R’ in Qimera).
After rejection, the point disappears from the display.

💡 Pro tip: Use the “3D view” to spot isolated spikes that are not obvious in 2D profiles. Spend most time on shallow areas (errors more visible).

5. Automatic Filters: Slope, Beam Pattern, Statistical

Automatic filters speed up cleaning. Common filters:

Slope filter: Removes points where the depth difference to neighbours exceeds a threshold (e.g., 20% of depth per beam).
Beam‑pattern filter: Rejects outer beams if their SNR is below a threshold.
Statistical filter (3‑sigma): Removes points deviating more than 3 standard deviations from the local median.
CUBE (Combined Uncertainty and Bathymetry Estimator): An advanced statistical method that flags outliers based on estimated uncertainty.

⚙️ Recommended workflow: Apply automatic filters first (conservative settings), then manually review flagged points. Never trust 100% automation.

🧹 Interactive Cleaning Simulator (Demo)

Click a button to simulate automatic removal of outliers from a sample profile.

Red profile with a spike (outlier). Click "Run Automatic Cleaning" to remove the spike.

6. Quality Control After Cleaning

After cleaning, perform these checks:

Cross‑line analysis: Compare cleaned main lines with cross‑lines. Residual difference should be < IHO tolerance (e.g., 0.15 m for shallow water).
Surface inspection: Generate a quick grid and look for unnatural steps or holes.
Swath overlap: Ensure that no artificial ridges appear at line edges.

✅ Good practice: Keep a copy of the uncleaned data and the cleaning flags separately. Never delete raw data.

7. Data Cleaning Checklist

Before cleaning: verify that tide and SVP corrections are applied.
Run automatic filters (slope, beam pattern) with conservative thresholds.
Manually review flagged points; reject genuine outliers.
Focus on shallow areas – noise is more critical there.
Check outer beams for SVP‑induced smile/frown; re‑cast SVP if needed.
Use cross‑line difference to validate cleaning.
Generate a surface and inspect for artefacts.
Save cleaning flags separately; do not modify raw files.
Document the cleaning parameters used.

Click items to track your progress (saved in browser).

8. Resources & Software Tools

.htmlCARIS HIPS那样SIPS with statistical and beam‑pattern filters那样teledynecaris.com.htmlHypack Processing那样Multibeam editing, cross‑line tool那样hypack.com.htmlMB‑System (open source)那样Command‑line cleaning, mbedit那样mbari.org

Software	Cleaning Features	Link
QPS Qimera那样Swath editor, CUBE, auto slope filter那样qps.nl/qimera

9. Frequently Asked Questions

How much cleaning is too much?

If you reject >5‑10% of soundings, revisit your acquisition. Excessive cleaning may remove real features.

Should I clean before or after tide/SVP corrections?

After applying corrections – otherwise outliers may be misidentified. Most processing software applies corrections first, then you clean.

Can I automatically remove all outer beams?

No, because outer beams contain valuable data. Only reject if beam pattern filter indicates low SNR.

What is the CUBE filter and why is it better?

CUBE (Combined Uncertainty and Bathymetry Estimator) uses statistical uncertainty propagation to identify outliers more robustly than simple slope filters. It is IHO‑recommended.

My cross‑line difference is still 0.3 m after cleaning – what now?

Check SVP, tide, and patch test again. A systematic difference indicates calibration error, not noise.

10. Action Items & Next Steps

📌 Open a sample MBES dataset in your processing software and explore the swath editor.
📌 Apply an automatic slope filter (e.g., 20% slope) and review the flagged points.
📌 Use the interactive simulator above to understand how outliers are removed.
📌 Proceed to Day 30: Surface Generation & Gridding.

🧹 Day 29: MBES Data Cleaning and Filtering Techniques