README.md

---
license: cc-by-4.0
language:
- en
pretty_name: Global Monthly PM2.5 from SatPM V6.GL.03
tags:
- geospatial
- timeseries
- air-quality
- pm25
- satellite
- remote-sensing
- netcdf
size_categories:
- n<1K
---
# 🌍 Global 0.01°-Resolution Monthly PM₂.₅ (1998–2024) — SatPM V6.GL.03

## Dataset Description

This repository is a Hugging Face mirror of the **SatPM V6.GL.03** high-resolution scientific dataset. It provides global monthly estimates of ground-level fine particulate matter concentration, PM₂.₅, at a fine spatial resolution of **0.01° × 0.01°**.

The estimates were generated by combining Aerosol Optical Depth (AOD) retrievals from multiple satellite-based instruments, including **MODIS, MISR, SeaWiFS, and VIIRS**, with information from the **GEOS-Chem** chemical transport model. These estimates were subsequently calibrated to global ground-based observations using a residual **Convolutional Neural Network (CNN)**.

The data are provided in gridded **NetCDF (.nc)** format using a WGS84 geographic coordinate system. This makes the dataset suitable for large-scale environmental health studies, air quality monitoring, epidemiological research, atmospheric science, environmental economics, and geospatial analysis.

The original data are natively hosted and maintained by **Washington University in St. Louis** through the SatPM project:

* Original source: [SatPM V6.GL.03 Dataset](https://www.satpm.org/v6-gl-03)

## Data Structure

The monthly high-resolution dataset is approximately **135.4 GB** in total size and is structured temporally.

```text
Monthly/
  V6GL03.CNNPM25.AF.199801-199801.nc
  V6GL03.CNNPM25.AF.199802-199802.nc
  ...
  V6GL03.CNNPM25.AF.202412-202412.nc
```

The repository contains **324 monthly NetCDF files**, covering the period from **January 1998 to December 2024**.

These files are useful for analyzing seasonal fluctuations, long-run air pollution trends, short-term air quality dynamics, and spatial variation in PM₂.₅ exposure at global scale.

## Spatial and Temporal Coverage

* Spatial coverage: global
* Spatial resolution: 0.01° × 0.01°
* Coordinate system: WGS84 geographic coordinates
* Temporal frequency: monthly
* Temporal coverage: January 1998 to December 2024
* Number of monthly files: 324
* File format: NetCDF `.nc`

## Important Resolution Note

Although the files are provided on a **0.01° grid**, users should be careful when interpreting the data at extremely local scales. Gridded products are provided to allow users to aggregate data to countries, regions, administrative units, watersheds, cities, or other spatial units of interest.

The estimates may not fully resolve PM₂.₅ gradients at the exact nominal grid resolution because the modeling framework incorporates information from coarser satellite, simulation, and monitoring sources.

## Main Variable

The main variable is monthly ground-level **PM₂.₅ concentration**, generally expressed in micrograms per cubic meter.

Users should inspect each NetCDF file directly for the exact variable name, units, dimensions, coordinate names, missing-value conventions, and file-level metadata.

## Intended Uses

This dataset is suitable for:

* global air pollution research;
* environmental exposure assessment;
* epidemiological and public health studies;
* climate and atmospheric composition research;
* environmental economics and development studies;
* spatial inequality and environmental justice research;
* country-level, region-level, or grid-level PM₂.₅ aggregation;
* long-run monthly panel construction;
* reproducible geospatial workflows in Python, R, Google Colab, and GIS software.

## Example Python Access

```python
from huggingface_hub import hf_hub_download
import xarray as xr

repo_id = "faviolc/monthlypm25"

path = hf_hub_download(
    repo_id=repo_id,
    repo_type="dataset",
    filename="Monthly/V6GL03.CNNPM25.AF.199801-199801.nc"
)

ds = xr.open_dataset(path)
print(ds)
```

## Example R Access

```r
library(ncdf4)

file_path <- "V6GL03.CNNPM25.AF.199801-199801.nc"

nc <- nc_open(file_path)
print(nc)
nc_close(nc)
```

## Recommended Workflow

Users can:

1. download one or more monthly NetCDF files;
2. load the files with `xarray`, `rasterio`, `terra`, `stars`, `ncdf4`, or other geospatial libraries;
3. subset the grid to a country, region, administrative boundary, watershed, or study area;
4. aggregate PM₂.₅ values to the desired spatial unit;
5. construct monthly, seasonal, or annual exposure panels.

## Attribution and Citation

This dataset is distributed under the **Creative Commons Attribution 4.0 International (CC BY 4.0)** license.

When using, transforming, or redistributing these data, users must provide appropriate credit to the original authors and link to the original SatPM project.

If you use this dataset in your research or projects, please cite the reference publication and the original data source.

### Reference Publication

Shen, S., Li, C., van Donkelaar, A., Jacobs, N., Wang, C., & Martin, R. V. (2024). **Enhancing Global Estimation of Fine Particulate Matter Concentrations by Including Geophysical a Priori Information in Deep Learning.** *ACS ES&T Air*.

DOI: [10.1021/acsestair.3c00054](https://doi.org/10.1021/acsestair.3c00054)

### Original Source

SatPM V6.GL.03 Dataset. Washington University in St. Louis.

Source: https://www.satpm.org/v6-gl-03

## License

The original SatPM V6.GL.03 product is distributed under the **Creative Commons Attribution 4.0 International (CC BY 4.0)** license. This Hugging Face mirror follows the same attribution requirements.

## Maintainer

This Hugging Face mirror is maintained by **Favio Leiva** to support reproducible research on air pollution, environmental exposure, public health, development, spatial inequality, and environmental change.