REDNET-ML: A Multi-Sensor Machine Learning Pipeline for Harmful Algal Bloom Risk Detection Along the Omani Coast

Ameer Alhashemi

cs.LGMar 4, 2026

REDNET-ML: A Multi-Sensor Machine Learning Pipeline for Harmful Algal Bloom Risk Detection Along the Omani Coast

Authors: Ameer Alhashemi

Organizations: School of Computer Science, University of Birmingham

Abstract

Harmful algal blooms (HABs) can threaten coastal infrastructure, fisheries, and desalination dependent water supplies. This project (REDNET-ML) develops a reproducible machine learning pipeline for HAB risk detection along the Omani coastline using multi sensor satellite data and non leaky evaluation. The system fuses (i) Sentinel-2 optical chips (high spatial resolution) processed into spectral indices and texture signals, (ii) MODIS Level-3 ocean color and thermal indicators, and (iii) learned image evidence from object detectors trained to highlight bloom like patterns. A compact decision fusion model (CatBoost) integrates these signals into a calibrated probability of HAB risk, which is then consumed by an end to end inference workflow and a risk field viewer that supports operational exploration by site (plant) and time. The report documents the motivation, related work, methodological choices (including label mining and strict split strategies), implementation details, and a critical evaluation using AUROC/AUPRC, confusion matrices, calibration curves, and drift analyses that quantify distribution shift in recent years.

Figures & tables

Figure 1: Study area and Sentinel-2 MGRS tiles used for the HAB fusion model, with plant AOIs (A–D) shown for plant-centric chipping and aggregation.

Figure 2: OCI-context composites used by REDNET-ML: MODIS-Aqua chlorophyll- a (left) and SST (right) over the Oman domain (2024 aggregate).

Decision	Implemented choice and rationale
Image evidence	Detectors provide score summaries (max/median/count) rather than final labels to improve robustness under sparse/noisy labels.
Detectors kept	Faster R-CNN (ResNet50) and SSD MobileNet retained due to strong region-level performance.
Detector excluded	YOLOv8 evaluated but excluded to reduce complexity (no material ensemble gain) [ 8 ] .
Fusion model	CatBoost decision fusion on MODIS, indices and detector scores [ 9 ] .
Operating metrics	Emphasis on AUPRC (primary) and AUROC (support) due to imbalance [ 16 ] .
Alerting policy	Two thresholds: $\tau_{\text{watch}}$ (candidate surfacing) and $\tau_{\text{action}}$ (higher-confidence escalation).

Table 1: Decision layer summary: model choices and how they are used at inference time.

Parameter	Range	Mean $\pm$ SD
chlor_a	0.044–9.972	0.312 $\pm$ 0.534
kd490	0.019–2.244	0.056 $\pm$ 0.072
nflh	-0.026–0.736	0.124 $\pm$ 0.109
sst	23.945–32.995	28.008 $\pm$ 1.853
fai_mean	-0.073–0.084	0.028 $\pm$ 0.025
ndwi_mean	-0.393–0.068	-0.156 $\pm$ 0.074

Table 2: Range of data used (training table): numerical ranges and mean

\pm

SD for key predictors.

Panel of Figure 3 — Figure 3: REDNET-ML architecture: multi-sensor evidence generation, non-leaky dataset construction, CatBoost decision fusion, and operational risk outputs.

Figure 4: REDNET HAB Ops Console (full GUI) used for operational exploration of plant risk, monthly context, and event-level evidence.

Figure 5: Pooled ROC and PR curves.

Model	Regional Recall	Mean Area Overlap	Takeaway
Faster R-CNN ResNet50	0.94	0.69	Best overall accuracy and localization
SSD MobileNet	0.94	0.66	Best efficiency–accuracy tradeoff
Faster R-CNN MobileNet	0.44	0.31	Capacity-limited; light baseline only
YOLOv8n3	0.92	0.50	Competitive recall; weaker alignment for this task

Table 3: Detector comparison using region-level metrics (

n=51

ground-truth regions).

Figure 6: Thresholded and calibration artifacts for the fusion model.

Figure 7: Model interpretability artifacts exported by the report scripts.

Remote Sensing

Remote sensing uses satellite, aerial, and ground-based observations to measure and map Earth and other planetary surfaces.

See all 528 papers

Predicting Pseudo-nitzschia harmful algal blooms along the Portuguese Coast using satellite-derived predictors
Jul 8, 2026Ayman Bnoussaad, El Khalil Cherif, Ligia Pinto +3Spatiotemporal Forecasting
ISR – Institute for Systems and Robotics, Lisbon, Portugal. · MARETEC/LARSyS, Associação do Instituto Superior Técnico para a Investigação e Desenvolvimento, Lisboa, Portugal. · IPMA – Portuguese Institute for Sea and Atmosphere, Algés, Portugal. +1
Landsat-Sentinel-2 Algal Bloom Mapping Using Vision Transformers: Model Description, Implementation, and Examples
Jun 15, 2026Thainara Lima, Vitor MartinsRemote Sensing Image Understanding Vision Transformer
a Department of Agricultural & Biological Engineering, Mississippi State University (MSU), Starkville, MS 39762, USA.
The impact of feature engineering and an optimisation framework for ocean colour machine learning
Aug 20, 2026Edson Silva, Julien Brajard, Simon Cappe +2Remote Sensing
Nansen Environmental and Remote Sensing Center, and Bjerknes Centre for Climate Research, Bergen, Vestland, Norway · Nansen Environmental and Remote Sensing Center, Bergen, Vestland, Norway

Jul 8, 2026J/K move · Enter open · S save

Predicting Pseudo-nitzschia harmful algal blooms along the Portuguese Coast using satellite-derived predictors

Ayman Bnoussaad, El Khalil Cherif, Ligia Pinto, Ramiro Neves+2

ISR – Institute for Systems and Robotics, Lisbon, Portugal. · MARETEC/LARSyS, Associação do Instituto Superior Técnico para a Investigação e Desenvolvimento, Lisboa, Portugal. · IPMA – Portuguese Institute for Sea and Atmosphere, Algés, Portugal.+1

Spatiotemporal Forecasting

Open paper AskPDF arXiv

Pseudo-nitzschia diatoms pose recurrent risks to coastal ecosystems and shellfish harvesting along the Portuguese Atlantic coast. Here we develop and evaluate a spatio-temporal machine-learning framework to predict harmful algal bloom (HAB) occurrence using exclusively satellite-derived predictors under realistic forecasting constraints. We characterised environmental and biological variability across shellfish production zones (L1-L9) using 5,882 observations, providing system-wide context. Predictive models were developed for zones L1-L2, a hotspot for Pseudo-nitzschia and domoic acid events, using a decade-long dataset (2013-2023; 1,440 observations; more than 1,000 satellite-based predictors including sea surface temperature, an upwelling index, chlorophyll-a, and plankton functional types). Sampling locations were partitioned into ecologically meaningful sub-regions using a river-aware spatial clustering scheme. A stringent spatio-temporal cross-validation strategy that simultaneously withholds entire years and spatial clusters prevents leakage and closely mimics real-world forecasting conditions. HAB occurrence proved moderately predictable across model classes and feature configurations. Ensemble tree-based methods achieved the strongest discrimination: Random Forest reached 0.74 +/- 0.05 with environmental predictors; Extra Trees reached 0.77 +/- 0.06 with biological variables added. Feature-importance analyses revealed that seasonal structure, spatial context, and lagged environmental conditions dominate model decisions, while biological indicators refine bloom likelihood within physically favourable periods. The framework demonstrates operationally relevant skill for satellite-supported HAB early-warning systems along eastern boundary upwelling coasts.

Abstract

Remote Sensing

Explore similar work

Predicting Pseudo-nitzschia harmful algal blooms along the Portuguese Coast using satellite-derived predictors

Landsat-Sentinel-2 Algal Bloom Mapping Using Vision Transformers: Model Description, Implementation, and Examples

The impact of feature engineering and an optimisation framework for ocean colour machine learning