Importance of Remote Sensing in Protecting Local Economies and Environments

Remote sensing
Environment
AI
Author

Hafez Ahmad

Published

October 16, 2022

Remote sensing is a vital tool for protecting the environment and supporting the livelihoods of coastal communities. By providing timely, large-scale, and high-resolution data on ocean color and water quality, remote sensing informs policies, guides resource management, and helps prevent economic and ecological disasters.

Fisheries and Aquaculture Monitoring

Remote sensing enables the monitoring of phytoplankton blooms, temperature anomalies, and turbidity, which directly impact fish health and productivity. This helps safeguard fisheries-dependent economies (Li et al., 2024).

This study also references the work of (Li et al., 2024), which discusses the prediction of daily distribution of chlorophyll-a using remote sensing, machine learning, and modeling techniques. This research is published in Science of the Total Environment and provides valuable insights into the integration of advanced technologies for environmental monitoring(Cen et al., 2022; Li et al., 2024; Shin et al., 2022).

Example (Gulf of Mexico): Remote sensing detected harmful algal blooms (HABs) in near-real time, allowing local fisheries to temporarily halt harvesting to avoid contaminated shellfish. This prevented health risks and long-term fishery bans(Ahmad & Jhara, 2025).

Tourism Protection

Beaches and coastal tourism areas are vulnerable to water pollution, HABs, and turbid plumes. Remote sensing provides early warning systems to maintain water quality standards.

Example (Florida, USA): Satellite detection of red tide events enabled state authorities to close affected beaches early, issue public advisories, and plan cleanup efforts. These actions minimized economic losses in the tourism sector.

Disaster Risk Reduction and Coastal Resilience

Satellite-derived data on sediment plumes, flood extent, and coastal erosion support disaster preparedness, helping communities recover faster and reduce losses.

Example (Bay of Bengal): After Cyclone Amphan (2020), remote sensing was used to map suspended sediment loads and salinity changes. These insights supported emergency freshwater supply efforts and aquaculture recovery.

AI in Ocean Color Remote Sensing

Ocean color imagery from satellites is widely used to monitor marine conditions, with chlorophyll‑a (Chl‑a) serving as a key indicator of phytoplankton biomass and primary productivity (Ahmad, Jose, Dash, Shoemaker, & Jhara, 2025).

AI and machine-learning methods are increasingly applied to derive water quality and biological variables from these data. Major applications include:

  • Chlorophyll Concentration Estimation: Machine-learning regression models (e.g., neural networks, random forests) map satellite reflectance to Chl‑a, extending traditional band-ratio algorithms. Chl‑a is crucial for tracking phytoplankton blooms and ocean productivity (Cen et al., 2022).

  • Harmful Algal Bloom (HAB) Detection/Forecasting: AI classifiers and time-series models identify and predict HAB events from ocean color. Approaches include CNN-based classification of bloom vs. non-bloom areas and time-series LSTM forecasting. Machine learning has successfully distinguished toxic red tides from other blooms in regional studies (Shin et al., 2022).

  • Turbidity and Suspended Matter Mapping: Regression models (e.g., random forest, SVR) use ocean color indices (e.g., TSS products) and ancillary data (wind, tide) to estimate turbidity extent. Forecasting turbid plumes one day ahead has been demonstrated by combining GOCI TSS maps with meteorological inputs.

  • Water Quality and Ecological Indices: Beyond Chl‑a, AI models infer composite indices (e.g., TRIX, eutrophication levels) from multispectral data. For example, random forests applied to Sentinel‑3/OLCI data can predict the Trophic Index (TRIX) with moderate success (R²≈0.37).

  • Data Fusion and Gap-Filling: Hybrid models combine process-based outputs (hydrodynamic/biogeochemical) with CNN or LSTM to produce high-resolution, gap-free Chl‑a fields. Such fusion captures both physical variability and optical signals, enabling reconstruction of daily Chl‑a distributions even during cloud cover.

Satellite and Sensor Data

Common ocean color sensors feeding AI models include:

  • SeaWiFS (1997–2010, NASA): Moderate resolution (~1 km) with 8 visible/near-IR bands. Provided global Chl‑a time series.

  • MODIS (Aqua/Terra, 2002–present, NASA): Visible and NIR bands (250 m–1 km) for daily coverage. Extensively used in ML models (e.g., HAB forecasting with daily composites).

  • VIIRS (SNPP/NOAA‑20, 2011–present): Successor to MODIS with 750 m resolution, blue-to-SWIR bands. NN algorithms using VIIRS M3–M5 (blue-green) and I1 (NIR) have improved Chl‑a retrievals in estuaries.

  • Sentinel‑3 OLCI (2016–present, ESA): Ocean and Land Colour Instrument with 300 m resolution and 21 bands. Provides high-quality Chl‑a and other water products. Used in CNN classification of blooms and in ML retrieval of ecological indices.

  • MERIS (Envisat, 2002–2012, ESA): 300 m, 15 bands. Legacy data still used in merged products.

  • Geostationary Imagers: GOCI (Korea, 2010–present): 6 visible bands (500 m, 1-hourly) over Asia, used for turbidity (TSS) mapping. GOCI TSS maps (490/745 nm) can drive ML turbidity forecasts.

  • High-Resolution and Hyperspectral: Sentinel‑2 MSI (10–60 m), Landsat OLI (30 m) and new hyperspectral sensors (e.g., PRISMA, CHIME) enable fine-scale water-quality mapping with ML. For instance, a Random Forest on 100+ band hyperspectral data (Zhuhai-1) achieved R²≈0.92 for Chl‑a in a hypersaline lake.

  • Multi-Sensor Merged Products: Platforms like Copernicus combine multiple sensors (SeaWiFS, MODIS, MERIS, VIIRS, OLCI) into gap-free daily Chl‑a composites at ~4 km, which can be ingested by AI models for consistent long-term mapping.

Satellite/Product Sensor/Instrument Spatial Resolution Temporal Resolution Key Applications
MODIS (Terra/Aqua) MODIS 250m – 1km Daily Ocean color, algal blooms, land cover, SST
Sentinel-2 MSI 10–60m 5 days Agriculture, forest cover, coastal zone monitoring
Sentinel-3 OLCI, SLSTR 300m (OLCI), 1km (SLSTR) 1–2 days Ocean color, SST, water quality, fire detection
Landsat 8 & 9 OLI, TIRS 30m (OLI), 100m (TIRS) 16 days Land use/cover change, coastal dynamics
VIIRS (Suomi NPP/NOAA-20) VIIRS 375m – 750m Daily Night lights, SST, ocean color
GPM GMI, DPR ~10km 2–3 hours (global revisit) Precipitation, flood early warning
ICESat-2 ATLAS Varies (laser altimetry) Seasonal to annual Coastal elevation, wetland dynamics, sea level change
PlanetScope RGB, NIR 3–5m Daily Precision agriculture, local monitoring, disaster response
SAR (e.g., Sentinel-1) C-band SAR 10–20m 6–12 days Flooding, mangroves, wetland changes, soil moisture

References

Ahmad, H., & Jhara, S. I. (2025). Mapping the dynamics of particulate organic carbon: Satellite observations of coastal to shelf variability in the northeastern gulf of mexico. Ocean Science Journal, 60(1), 7.
Ahmad, H., Jose, F., Dash, P., Shoemaker, D. J., & Jhara, S. I. (2025). Machine learning-based estimation of chlorophyll-a in the mississippi sound using landsat and ocean optics data. Environmental Earth Sciences, 84(7), 172.
Cen, H. et al. (2022). Applying deep learning in the prediction of chlorophyll-a in the east china sea. Remote Sensing, 14(21), 5461. https://doi.org/10.3390/rs14215461
Li, H. et al. (2024). Prediction on daily distribution of chl-a via remote sensing + ML + modeling. Science of The Total Environment, 910, 168642. https://doi.org/10.1016/j.scitotenv.2023.168642
Shin, J. et al. (2022). CNN model for discrimination of HAB from non-HABs using sentinel-3 OLCI imagery. ISPRS Journal of Photogrammetry and Remote Sensing, 191, 250–262. https://doi.org/10.1016/j.isprsjprs.2022.07.012