Multispectral Imaging for Caulerpa Detection and Condition Monitoring

Early multispectral trials with NIWA using BioSync’s field-ready detection workflow

Background: why detection is the bottleneck

Invasive Caulerpa can establish quickly, spread patchily, and present in visually complex habitats where colour and texture cues are inconsistent. In practice, survey teams often face low contrast imagery, mixed substrates, variable lighting, and water-column effects that obscure early-stage growth. These conditions make consistent identification difficult from standard colour imagery alone, particularly when the objective is not merely to “spot it once,” but to monitor change over time in a way that is repeatable, auditable, and operationally scalable.

BioSync is being developed to support this monitoring problem: converting underwater imagery into structured outputs that help teams locate likely Caulerpa, compare conditions across time, and produce evidence-ready reporting products.

Why multispectral helps 

Multispectral imaging extends beyond standard visible colour channels by capturing additional spectral information that is not available in conventional RGB video. For benthic algae, the key advantage is not aesthetic—rather, it is analytical: reflectance and absorption characteristics can diverge in ways that are subtle (or invisible) in RGB but more separable when additional spectral channels are available.

In practice, multispectral data can improve signal discrimination under variable illumination and background complexity by enabling:

• Spectral contrast views that reduce reliance on colour alone

• Derived index-style representations (heatmaps) that emphasise biologically relevant differences

• More repeatable segmentation candidates (regions of interest) suitable for time-series comparison

The goal is not to replace expert judgement, but to reduce ambiguity and improve consistency when screening imagery and documenting change.

Adding chlorophyll context: from “what is it?” to “what condition is it in?”

Beyond presence/absence detection, biosecurity teams often need to understand condition and change for example, whether a patch appears stable, expanding, or responding to a management intervention. One promising direction is leveraging multispectral information to infer pigment-related signatures, including signals associated with chlorophyll (Chl).

Chlorophyll is a core photosynthetic pigment, and changes in chlorophyll-related signatures can provide a non-invasive proxy for changes in algal condition. In practical terms, this can support:

• Condition monitoring: observing shifts that may correlate with stress, decline, or recovery

• Time-series comparison: detecting subtle changes that may not be obvious in RGB

• Treatment response screening: providing an additional lens to evaluate whether a patch is changing following intervention

BioSync’s approach treats chlorophyll-related outputs as decision support, not a definitive biochemical measurement. The intent is to provide interpretable views and summaries that help prioritise review and field verification, while maintaining auditability of the underlying imagery.

Controlled trials with NIWA: what we tested

BioSync has been trialled in controlled conditions at NIWA to evaluate whether multispectral imagery can support reliable, repeatable identification of Caulerpa and provide interpretable condition cues in practical survey-like settings. Controlled trials are valuable because they allow key variables to be explored systematically—such as lighting conditions, camera angle, distance to target, background substrate, and water clarity—while maintaining a known scene composition.

The focus of these trials was methodological validation at a high level:

1. Image acquisition feasibility: obtaining stable, interpretable multispectral and visible imagery under conditions relevant to field deployments

2. Separability: assessing whether Caulerpa exhibits consistent contrast in multispectral-derived views relative to surrounding substrates and non-target material

3. Operational outputs: ensuring results can be expressed as products useful to monitoring teams (overlays, candidate regions, summary statistics), rather than remaining a purely research-grade dataset

4. Condition cues: exploring whether pigment-linked representations (including chlorophyll-related views) provide usable signals for monitoring change

Importantly, this work is framed around operational monitoring needs: repeatability, interpretability, and exportable outputs suitable for evidence-based reporting.

---

Caption (Figure 1): BioSync prototype interface showing synchronized multispectral and visible views, candidate regions of interest, and derived heatmap representations designed to support repeatable screening and monitoring.

What the BioSync workflow produces

BioSync is being designed as a decision-support layer that sits between raw imagery and field reporting. Instead of requiring users to manually interpret long video sequences, the workflow aims to summarise imagery into structured, reviewable outputs. In its current form, outputs are oriented around five practical deliverables:

1) Detection overlays
Candidate regions are identified and presented as overlays on the visible scene. This supports rapid screening and enables reviewers to focus effort where evidence is strongest.

2) Heatmap-style representations
Multispectral-derived heatmaps help visualise contrast patterns that may indicate Caulerpa presence or changes in condition. These representations are intended to be interpretable by field teams without requiring specialist knowledge of the underlying computation.

3) Chlorophyll views (condition support)
BioSync can generate additional representations designed to reflect pigment-related changes consistent with chlorophyll signals. Used carefully, these can assist with monitoring workflows where teams need to assess whether a patch is changing through time.

4) Region-of-interest (ROI) tracking
ROIs can be compared across time or across repeated passes, supporting monitoring workflows where the objective is not only detection, but change detection (e.g., growth, decline, disturbance, treatment response).

5) Evidence-ready exports
Outputs can be assembled into evidence packs: timestamped frames, annotated imagery, and concise summaries that can be incorporated into routine reporting. The emphasis is on clarity and auditability—critical for biosecurity response and inter-agency coordination.

This approach prioritises operational value: faster review, consistent interpretation, and outputs that can be shared across teams.

Field deployment considerations: making it practical

A central design constraint for BioSync is that it must function in the real world—on vessels, in mobile setups, and with practical limitations on power, compute, and operator time. As a result, the system is being developed as a portable workflow that can run in field environments and integrate with typical acquisition platforms (e.g., towed cameras, ROV video streams, fixed stations).

Caption (Figure 2): A field-oriented BioSync setup designed for practical deployment, supporting real-time review and structured outputs from underwater imagery.

This field emphasis matters because biosecurity monitoring frequently depends on:

• consistent procedures across multiple sites

• repeatable outputs suitable for long-term trend comparison

• rapid reporting during response phases

Next steps

The NIWA trials represent an early validation phase focused on feasibility, interpretability, and operational outputs. Next stages will broaden testing to additional conditions and deployment contexts, including more complex backgrounds and operational acquisition platforms. BioSync is being developed with the intention of supporting scalable monitoring workflows where data products can be exported into mapping and reporting systems used by agencies and contractors.

Collaboration and deployment interest

BioSync is actively seeking collaboration opportunities where multispectral-assisted monitoring could accelerate detection, reduce uncertainty, or improve repeatability of Caulerpa assessments. This includes partners with sites of interest, existing camera workflows, monitoring programmes, or operational needs for evidence-ready reporting.