README.md

---
license: cc-by-4.0
tags:
- earth-observation
- remote-sensing
- satellite
- geospatial
- night-lights
configs:
- config_name: default
  data_files:
  - split: train
    path: INDEX.parquet
---


![darktom_1](https://cdn-uploads.huggingface.co/production/uploads/6304c06eeb6d777a838eab63/BURIJVY40Q2IxBjWt-1vt.png)

# Major TOM Core VIIRS Nighttime Light
Annual radiance composites from the Visible Infrared Imaging Radiometer Suite (VIIRS) Day/Night Band (DNB), packaged on the Major TOM grid.

| Source | Modality | Number of Patches | Patch Size | Total Pixels |
|:-------|:--------:|:-----------------:|:----------:|:------------:|
| VIIRS Nighttime Light | Nighttime Light | 2,108,341 | 1056 × 1056 (10 m) | > 2.3 trillion |

**Authors:** Marcin Kluczek ([CloudFerro](https://cloudferro.com)), Mikolaj Czerkawski ([Asterisk Labs](https://asterisk.coop)), Cesar Aybar ([Asterisk Labs](https://asterisk.coop)), Jędrzej S. Bojanowski ([CloudFerro](https://cloudferro.com))

# Content

| Variable | Description | Units |
|----------|-------------|-------|
| `median` | Median annual radiance | nW/cm²/sr |

> Additional modalities (`average`, `average-masked`, `median-masked`, `min`, `max`, `cf_cvg`, `cvg`) will be supplied later.

Source product:
[Annual VIIRS Nighttime Lights V2](https://eogdata.mines.edu/products/vnl/)
- 2016–2021: Annual VNL V2.1
- 2022–2024: Annual VNL V2.2

Per-year patch counts:

| Year | Patches |
|-----:|--------:|
| 2016 | 55,780 |
| 2017 | 20,541 |
| 2018 | 85,109 |
| 2019 | 412,181 |
| 2020 | 384,748 |
| 2021 | 378,299 |
| 2022 | 422,468 |
| 2023 | 348,773 |
| 2024 | 442 |

# Spatial Coverage

Global, on the Major TOM 10 m grid, spanning approximately 75°N to 65°S in latitude and the full 180°W to 180°E longitude range. This is a global monotemporal dataset: nearly every land patch is contained at least once per year, with only marginal overlaps.

# Cloud Optimized ZIP (cozip)

This dataset is published as **cozip** archives — standard ZIP files arranged so that any individual patch can be fetched in a single HTTP range request, without downloading the whole shard. cozip is developed by [Asterisk Labs](https://github.com/asterisk-labs/taco/tree/main/cozip).

**Layout on disk:**
```
INDEX.parquet                                  # global catalog of all patches
2016/MAJORTOM-VIIRS-NTL_2016_median_000.zip    # cozip shards (~4 GiB each)
2016/MAJORTOM-VIIRS-NTL_2016_median_001.zip
...
2024/MAJORTOM-VIIRS-NTL_2024_median_000.zip
```

**`INDEX.parquet`** is a GeoParquet catalog with one row per patch. Key columns:

| Column | Meaning |
|--------|---------|
| `name` | Patch filename (`MT10_<col>D_<row>L_<year>_median.tif`) |
| `shard` | Relative path to the cozip shard that holds this patch |
| `offset`, `size` | Byte range inside the shard — the raw GeoTIFF payload |
| `id`, `majortom:code_100km`, `majortom:code_1000km` | Major TOM grid identifiers |
| `geometry`, `bbox` | Patch footprint (WGS84) |
| `admin:country/state/district` | Administrative attribution |
| `terrain:elevation`, `climate:*`, `soil:*`, `socio:*` | Co-located covariates |

## Quick start

Install the reader:
```bash
pip install cozip rasterio pandas pyarrow
```

**Read the global catalog** and stream one patch directly from the Hub — no full-shard download:

```python
import cozip
import rasterio

# Read INDEX.parquet from anywhere (local path or https URL).
df = cozip.read("https://huggingface.co/datasets/Major-TOM/Core-VIIRS-Nighttime-Light/resolve/main/INDEX.parquet")

# Pick a patch by id (or filter by bbox, country, year, etc.).
row = df[df["id"] == "MT10_696D_231L"].iloc[0]

# `cozip:gdal_vsi` is a /vsisubfile/ path GDAL can open directly.
with rasterio.open(row["cozip:gdal_vsi"]) as src:
    arr = src.read(1)
    print(arr.shape, src.crs, src.bounds)
```

**Read a single shard's metadata only** (when you already know which shard you want):

```python
df = cozip.read("https://huggingface.co/datasets/Major-TOM/Core-VIIRS-Nighttime-Light/resolve/main/2020/MAJORTOM-VIIRS-NTL_2020_median_000.zip")
print(len(df), "patches in this shard")
```

**Pure-Python access** (no GDAL, just an HTTP Range request):

```python
import requests, io, rasterio

url = f"https://huggingface.co/datasets/Major-TOM/Core-VIIRS-Nighttime-Light/resolve/main/{row['shard']}"
off, sz = int(row["offset"]), int(row["size"])
blob = requests.get(url, headers={"Range": f"bytes={off}-{off + sz - 1}"}).content
with rasterio.io.MemoryFile(blob) as mem, mem.open() as src:
    arr = src.read(1)
```

# Cite
[![arxiv](https://img.shields.io/badge/Open_Access-arxiv:2402.12095-b31b1b)](https://arxiv.org/abs/2402.12095/)
```latex
@inproceedings{Major_TOM,
  title={Major TOM: Expandable Datasets for Earth Observation},
  author={Alistair Francis and Mikolaj Czerkawski},
  year={2024},
  booktitle={IGARSS 2024 - 2024 IEEE International Geoscience and Remote Sensing Symposium},
  eprint={2402.12095},
  archivePrefix={arXiv},
  primaryClass={cs.CV}
}
```

# VIIRS Nighttime Light products credit

Many thanks to [**Earth Observation Group (EOG)**](https://payneinstitute.mines.edu/eog/), part of the Payne Institute for Public Policy at Colorado School of Mines, especially **Christopher D. Elvidge**, for their support and consultation.

```
- C.D. Elvidge, M. Zhizhin, T. Ghosh, F-C. Hsu, "Annual time series of global VIIRS nighttime lights derived from monthly averages: 2012 to 2019", Remote Sensing, 2021, 13(5), 922. https://doi.org/10.3390/rs13050922.
- C.D. Elvidge, K. Baugh, M. Zhizhin, F.-C. Hsu, and T. Ghosh, "VIIRS night-time lights," International Journal of Remote Sensing, vol. 38, pp. 5860–5879, 2017. https://doi.org/10.1080/01431161.2017.1342050.
- C.D. Elvidge, K.E. Baugh, M. Zhizhin, and F.-C. Hsu, "Why VIIRS data are superior to DMSP for mapping nighttime lights," Asia-Pacific Advanced Network 35, vol. 35, p. 62, 2013. http://dx.doi.org/10.7125/APAN.35.7.
```