UBC & MSI Research Projects

Jun 11, 2025 | Research

Herring Spawn and Marine Debris Trends in Atl’ka7tsem /Howe Sound

UBC Master’s Graduates Eron Macartney and Tyler Zhang analysed data from the Marine Reference Guide, helping to inform future Marine Stewardship Initiative projects.

Left: Eron Macartney presenting his project Predicting Pacific Herring Spawn Events in the Howe Sound, British Columbia. Right: Tyler Zhang presenting his project Marine Debris Hotspot Analysis in Howe Sound, British Columbia

The Master of Geomatics for Environmental Management (MGEM) program at the University of British Columbia is a professional master’s degree that trains students to use geospatial tools to inform the management and regulation of natural areas and resources. Students partner with industry and environmental organizations to work on research objectives with real-world data.

Two students from the 2024/2025 MGEM program, Eron Macartney and Tyler Zhang, worked with data from the MSI’s Marine Reference Guide to explore patterns in biological and physical processes in Átl’ka7tsem / Howe Sound.

Below is a summary of the two projects as well as links to the full reports.

Marine Debris Hotspot Analysis in Howe Sound, British Columbia

Tyler (Dongping) Zhang
Master of Geomatics for Environmental Management

Marine debris threatens marine ecosystems, infrastructure, and public health. Monitoring of marine debris in Howe Sound is limited due to the fjord’s semi-enclosed geography and uneven population distribution. Reporting of debris is necessary for management, but public reports are biased towards more populated areas. This study aimed to:

  • Identify landcover features common to known debris hotspots using Sentinel-2 satellite imagery.

  • Simulate the movement of floating debris using river discharge and seasonal wind data.

  • Predict and validate new debris accumulation hotspots.

 

Read the full report

Methods:

 

1. Supervised Classification of Satellite Imagery

To understand the relationship between debris hotspot locations and land cover types in Howe Sound, the study used multi-band Sentinel-2 satellite imagery from both summer and winter and applied supervised classification to classify landcover into landcover classes. Supervised classification is a machine learning process where the user “teaches” the computer to predict or identify data based on training examples provided by the user.

 

  • Six landcover classes were identified: vegetation, ocean, urban, river discharge, snow, and rock.

  • Tested three classification algorithms: Maximum Likelihood (MLC), Random Forest, and Neural Network.

  • The Neural Network classifier yielded the highest accuracy (0.98 in summer, 0.97 in winter, with 1 being the highest possible accuracy).

    Left: Landcover classification of Howe Sound in the summer, with red dots representing reported marine debris.

    Right: Landcover classification of Howe Sound in the winter, with red dots representing reported marine debris.

    2. Debris Transport Simulation

    Using historical wind direction data, average wind directions for summer and winter were calculated and used to train a simulator that predicts the movement of marine debris throughout the sound in those respective seasons.

    • Simulated debris movement using wind and river vectors across three fjord sections: Northern, Central, and Southern Howe Sound.

    • Wind data from 4 weather stations (Squamish Airport, Pam Rock, Port Mellon, Point Atkinson) processed into wind rose diagrams.

    • River discharge from Squamish River modeled using a fan-spread function to reflect how freshwater flows affect surface transport.

    Results:

    49 hotspots were predicted:

    • 20 matched existing reports from the MSI’s Marine Debris Reporting Tool.
    • 29 were new, including 10 along the western shoreline—a historically under-monitored area.

    • Classification revealed that many debris hotspots occur near urban shorelines and semi-enclosed bays —locations prone to trapping floating material.

    • Seasonal winds strongly influenced debris direction: for instance, easterly winter winds in the south vs. southerly summer inflows

    Read the full report

    Right: Reported marine debris in red, predicted marine debris in yellow.

    Predicting Herring Spawn in Howe Sound, British Columbia

    Eron Macartney
    Master of Geomatics for Environmental Management

    The timing and distribution of herring (Clupea pallasii) in Howe Sound has not been well documented historically, and the environmental drivers within the sound that influence these are not well understood.  It is generally understood that the environmental factors of sea surface temperature (SST), photoperiod, and lunar phase influence the timing of spawn events. Using four years of MSI herring spawn data (2021-2024) the study aimed to explore:

     

    • What is the influence of sea surface temperature, photoperiod, and lunar cycles on the spawn timing of Pacific herring in Átl’ka7tsem?
    • Is there a significant relationship between these environmental factors and spawn timing?
    Read the full report

    Methods: 

     

    1. Data Collection

    • Spawn dates and locations recorded by the Marine Stewardship Initiative (MSI) from 16 sites via foot, snorkel, and boat.

    • Environmental data included:

      • SST from Landsat 8/9 (Band 10 TIR), processed using ArcGIS Pro and R.

      • Lunar phases from R’s lunar package.

      • Photoperiod (hours of daylight) calculated using the RStudio package suncalc

    2. Statistical Analysis

    • Ran logistic regression (General Linear Model), chi-square tests, point-biserial correlations, and a random forest model.

    • Compared spawn/no-spawn binary outcomes with environmental predictors.

     Left top: Recorded spawn events by MSI staff and volunteers, organized by lunar phase at time of spawn.

    Left bottom: Predicted spawn events based on previous years data and lunar phase.

    Results

     

    • Lunar Phases were the strongest predictor.

      • Waxing Crescent phase showed a 36% probability of spawning (p < 0.05).

      • Full and First Quarter moons were also influential (p < 0.10).

     

    • Sea Surface Temperature (SST) and photoperiod showed no significant correlation.

      • Most spawn events occurred near average SST, but colder deviations reduced spawn likelihood.

      • Photoperiods during spawning ranged from 11–13 hours of daylight, but correlation was insignificant (p = 0.942).

    Read the full report

    These studies provide insight that helps to inform future MSI projects, such as marine debris clean-ups and improving survey methodology in the annual Searching for Slhawt’ / Herring Spawn Survey Program.

    Thank you Tyler and Eron for your hard work, and thank you to the UBC MGEM teaching staff and coordinator for supporting the students in their projects.

    If you have ideas for studies involving data from the Marine Reference Guide, please feel free to connect with us anytime.