Network panel

Algae Dataset

Every data hub, dataset, sensor, variable, paper and limitation in one tree — and for each: where it comes from, how far back it goes, how often it captures, why it belongs in a cholera study, and why it does not work everywhere.

Highlight site All three Chakaria HDSS Matlab HDSS Dhaka City Choosing a site greys out everything that does not apply there.

TEMPORAL limitations

CRITICAL Record starts mid-2015 - two-thirds of the study window is uncovered temporal

Sentinel-2 L1C begins June 2015 and L2A March 2017. The study period starts in 2000, so the single most capable inland sensor is absent for 15 of 26 years.

What to do: Landsat 30 m is the only backfill. Accept a 16-day cadence before 2015, bias-correct across the boundary, and consider modelling the pre-2015 era monthly.

CRITICAL OCTS and CZCS predate the study period temporal blocks Chakaria blocks Matlab blocks Dhaka

CZCS ends 1986 and OCTS ends 1997. Neither overlaps 2000-2026 meaningfully.

What to do: Not used. Listed for completeness only.

HIGH The 2015 sensor discontinuity temporal

Sentinel-2 begins mid-2015 and Sentinel-3 late 2016. Before that you have Landsat at 16 days and MODIS at 250 m. Blending sensors without correction creates a step change in 2015 that a model will read as a trend - a completely artefactual one.

What to do: 11_algae_daily_fill.py regresses every sensor onto a reference on same-day overlaps and writes the coefficients to algae_daily_bias_coeffs.csv. Any pair with fewer than 20 overlaps is left UNCORRECTED and flagged rather than silently fudged.

HIGH 16-day revisit misses short blooms entirely temporal

One Landsat satellite passes every 16 days. A bloom that initiates, peaks and collapses inside two weeks can fall entirely between two passes and leave no trace in the record.

What to do: Two satellites in orbit halve the effective interval to about 8 days. Combine L7+L5 for 2000-2012 and L8+L9 for 2021 onward. Report bloom FREQUENCY with an explicit detection-probability caveat.

HIGH Mission ended December 2010 temporal

SeaWiFS stops on 2010-12-11, so it covers only the first 11 years of the study window.

What to do: Use it for 1997-2010 and hand over to MODIS-Aqua or the merged OC-CCI record, which is already inter-calibrated across sensors.

MEDIUM Monthly cadence cannot resolve a two-week bloom temporal

OC-CCI monthly is the defensible cadence for this product, but a bloom that initiates and collapses inside four weeks is invisible in it.

What to do: Pair with MODIS 8-day or VIIRS daily where the coastal signal needs finer timing - accepting that those are noisier.

MEDIUM PACE launched 2024 - a very short record temporal

Best pigment discrimination available anywhere, and only about two years of data inside the study window.

What to do: Not usable for the time-series analysis. Worth one figure as a demonstration of what hyperspectral adds.