PranavKonijeti/viirs-sst-daily-nonNRT

VIIRS/SNPP Daily Sea Surface Temperature — 4 km, Science Quality

Daily global sea surface temperature from the VIIRS instrument aboard Suomi-NPP, Level-3 Standard Mapped Image at 4 km, as distributed by the NASA Ocean Biology Processing Group. This is the science-quality (refined) feed — fully calibrated and reprocessed.

A companion repository, `PranavKonijeti/viirs-sst-daily-nrt`, holds the near-real-time (NRT) feed of the same product. See Which feed should I use? below.

At a glance

Files

One netCDF-4 file per day, named:

SNPP_VIIRS.YYYYMMDD.L3m.DAY.SST.sst.4km.nc

Each file contains:

Land, cloud, and retrieval failures are NaN after decoding. Typically ~70–85% of the grid is NaN on any given day — most of the planet is either land or under cloud, so expect sparse coverage per file and use multi-day composites if you need dense fields.

Usage

from huggingface_hub import snapshot_download

path = snapshot_download(
    repo_id="PranavKonijeti/viirs-sst-daily-nonNRT",
    repo_type="dataset",
    local_dir="./sst_non-nrt",
)

To pull a single year instead of all 63 GB:

snapshot_download(
    repo_id="PranavKonijeti/viirs-sst-daily-nonNRT",
    repo_type="dataset",
    allow_patterns="SNPP_VIIRS.2015*.nc",
    local_dir="./sst_2015",
)

Reading one day:

import xarray as xr

ds = xr.open_dataset("SNPP_VIIRS.20150104.L3m.DAY.SST.sst.4km.nc")
sst = ds["sst"]            # (lat, lon), degrees Celsius, NaN where no retrieval

Two things worth knowing when reading these:

• —sst is stored as scaled int16. xarray decodes it to floating-point Celsius automatically; if you read with a lower-level netCDF library, apply scale_factor and add_offset yourself.
• —Some L3m products in this family are distributed in Kelvin. These files are degree_C, but if you mix in other sources, a mean > 100 check is a reliable way to catch a Kelvin array that slipped through.

Which feed should I use?

Use this repository for anything scientific. The NRT feed exists to be fast, not correct, and its calibration can shift retroactively as granules are reprocessed — which makes it a poor basis for a model or a trend estimate. The two overlap from October 2019 onward, so the same date may appear in both with slightly different values. That is expected, not an error.

Gaps and missing data

Days are not contiguous. 199 of the 5,247 days in the span have no file, from instrument downtime, processing gaps, or granules that failed download. Do not assume file n+1 is the day after file n — parse the date from the filename.

Coverage is also thinner in the early part of the record. Within any given file, per-pixel gaps from cloud and land are the norm rather than the exception.

Provenance

Downloaded from the NASA Ocean Biology Processing Group's distribution server at oceandata.sci.gsfc.nasa.gov. Files are unmodified as distributed — no resampling, reprojection, or filtering was applied.

• —Platform: Suomi-NPP · Instrument: VIIRS · Processing level: L3 Mapped
• —Producer: NASA/GSFC/OBPG
• —Title as distributed: VIIRSN Level-3 Standard Mapped Image

This repository is a convenience mirror. For authoritative, current, or differently processed versions of this product, go to the source above.

License and citation

Distributed under the NASA Earth Science Data and Information Policy, as recorded in each file's license global attribute. NASA Earth science data are open; attribution to NASA/GSFC/OBPG is expected.

If you use this data, cite the NASA Ocean Biology Processing Group rather than this mirror:

NASA Ocean Biology Processing Group. Visible and Infrared Imager/Radiometer Suite (VIIRS) Level-3 Mapped Sea Surface Temperature Data, Suomi-NPP. NASA Ocean Biology Distributed Active Archive Center. https://oceandata.sci.gsfc.nasa.gov

Notes

The dataset viewer is not available here — HF renders Parquet and similar tabular formats, and these are raw netCDF grids. Use xarray as shown above.