Chlorophyll & Visibility: Reading Ocean Clarity from Plankton Levels

2026-03-16

Chlorophyll concentration indicates the amount of phytoplankton in seawater. The common assumption is that more plankton means greener, more turbid water — but our data analysis reveals a more nuanced picture. By matching 46,000+ dive logs with satellite data, we uncover the true relationship between chlorophyll and visibility.

15.2m

Low chlorophyll

13.2m

Mid chlorophyll

14.7m

High chlorophyll(!)

Average Visibility by Chlorophyll Level

Chlorophyll LevelAvg Vis (m)
Low (< 0.5 mg/m³)15.2m
Mid (0.5–2 mg/m³)13.2m
High (> 2 mg/m³)14.7m
Note: High chlorophyll (> 2 mg/m³) waters show higher visibility than mid-range — a surprising reversal.

The 'High Chlorophyll but Clear' Paradox

Cause: Confounding site characteristics

The main cause is that inherently clear sites like Yonaguni, due to geographical characteristics (Kuroshio influence, upwelling), show relatively high satellite-measured chlorophyll. Yonaguni averages 24.5m visibility (top-ranked) while also showing elevated chlorophyll levels.

Statistical confounding

When trying to predict visibility from chlorophyll alone, confounding factors like site geography (latitude, oceanic exposure, bathymetry) interfere. This is why the simple chlorophyll-to-visibility correlation isn't as strong as Kd490's.

Kd490 Comparison: Why Kd490 Predicts Better

Among satellite data, Kd490 is a stronger visibility predictor than chlorophyll. Kd490 ranks #3 in AI model importance, while chlorophyll ranks lower.

MetricBest vs Worst GapAI ImportanceMonotonic?
Kd49012.0m#3Yes
Chlorophyll2.0mLowerNo

Kd490 directly measures light attenuation, giving it a clear causal link to visibility. Chlorophyll is an indirect proxy (plankton -> turbidity -> reduced visibility), where confounding factors can intervene.

When Chlorophyll IS Useful

Temporal changes at the same site

The confounding issue arises in 'cross-site comparison.' Within the same site's time series, chlorophyll spikes effectively indicate spring bloom or red tide. For example, the Izu Peninsula shows clear synchronization between spring chlorophyll rises and visibility drops.

Supporting role in AI models

The AI model uses chlorophyll combined with other features. While weak alone, it contributes to prediction when combined with site and seasonal information.

Summary: Choosing the Right Satellite Metric

  • Pre-dive visibility check -> Prioritize Kd490
  • Spring bloom / red tide early warning -> Chlorophyll is effective
  • Most accurate prediction -> Use AI model that includes both

About the Data

Chlorophyll data sourced from NOAA ERDDAP (VIIRS/SNPP satellite). Visibility data uses 46,000+ real dive log observations. Chlorophyll is log-transformed (chlorophyll_log) as AI model input.

Reference: NASA Ocean Biology Processing Group (https://oceancolor.gsfc.nasa.gov/)

🌊 Check Visibility Forecasts

View AI-powered 7-day visibility forecasts for 30+ dive sites across Japan.

Open Forecast App →