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.

ACCESS limitations

CRITICAL Outcome data is behind icddr,b ethical approval access

Cholera case series for all three sites require Research Review Committee and Ethical Review Committee approval plus a data-sharing agreement. Free for collaborators, but it takes weeks to months. Every satellite task can run in parallel · none of them can finish without it.

What to do: Submit the application before starting anything else. Ask for three things in one email: the Matlab village-ID (vid) lookup, the Chakaria union-level shapefile, and whether Chakaria records lab-confirmed cholera or only diarrhoeal disease.

CRITICAL Both daily VIIRS ERDDAP endpoints are dead from this network access

00_verify_sources.py recorded TIMEOUT on noaacwNPPVIIRSchlaDaily and noaacwNPPN20VIIRSchlociDaily. That removes the easiest daily ocean-colour route entirely.

What to do: Reach VIIRS through Google Earth Engine instead. 12_download_all_free.py retries with the latitude axis flipped, logs the failure and carries on rather than crashing.

CRITICAL icddr,b environmental archive needs IRB approval access

The Huq et al. 2005 paired pond dataset - V. cholerae presence with chlorophyll and temperature, sampled fortnightly - is the highest-value validation source in the entire catalogue and it sits behind ethical approval.

What to do: Include it in the same data request as the outcome data. It costs nothing extra to ask.

CRITICAL Matlab village ID key is effectively empty access blocks Matlab

The vid field is populated in 8 of 251 records, and 251 polygons carry only 245 distinct village names with six names used twice. Neither an ID join nor a name join to the HDSS outcome data is safe.

What to do: Request the village-ID lookup table from icddr,b. Until it arrives, model at BLOCK level - seven units - and extract at village level anyway so nothing has to be redone later.

HIGH ACOLITE is roughly 1 GB per scene and weeks of processing access

The rigorous route to real reflectance requires downloading L1C scenes and running Dark Spectrum Fitting locally. For a multi-year archive across three sites this is a serious computing commitment.

What to do: Do it AFTER the ordinal pipeline works end to end. Filter to under 20 percent cloud first, and consider reprocessing only the bloom-season scenes.

MEDIUM Needs a Copernicus Marine account and a Python toolbox access

Free, but registration plus pip install copernicusmarine plus a login step, and the output is NetCDF requiring conversion.

What to do: Worth it - this is the only daily gap-free chlorophyll product available. Convert with xarray and netCDF4.