Skip to content

Global Annual Simulated NPP-VIIRS Nighttime Light Dataset (1992-2023)

The SVNL (Simulated VIIRS Nighttime Light) dataset provides a harmonized, continuous global annual NPP-VIIRS-like nighttime light data from 1992 to 2023. The longest available consistent NPP-VIIRS-like nighttime light dataset spanning over 32 years through cross-sensor calibration and deep learning methods.This dataset addresses critical gaps in existing nighttime light data by bridging DMSP-OLS (1992-2013) and NPP-VIIRS (2012-2023) through an innovative deep learning approach using a Nighttime Light U-Net super-resolution network (NTLSRU-Net).

The dataset features temporal coverage from 1992-2023 (32 years) with a spatial resolution of 15 arc-seconds (~500m). It provides global coverage from -180° to 180° longitude and 65°S to 75°N latitude. The data maintains consistent calibration across DMSP-OLS and VIIRS sensors while providing enhanced spatial detail comparable to NPP-VIIRS and reduced saturation and blooming effects compared to raw DMSP data. The dataset has been extensively validated against socioeconomic indicators and existing nighttime light datasets.

The dataset enables long-term analysis of urbanization monitoring, socioeconomic estimation, environmental assessment, human activity patterns, and regional and global development studies. This comprehensive temporal coverage makes it particularly valuable for studying long-term urban development, economic growth patterns, and environmental changes at both regional and global scales. You can read more in the paper here and find the dataset here

Methods

The dataset was generated through: 1. Preprocessing of raw DMSP-OLS data to address interannual inconsistency 2. Cross-sensor calibration using NTLSRU-Net with Landsat NDVI assistance 3. Integration of simulated VIIRS data (1992-2011) with real VIIRS data (2012-2023) 4. Comprehensive validation across multiple scales and metrics

Dataset Preprocessing

Original data was not compression optimized , an LZW compression with tiling was applied to convert these into COG for direct use as well as being ingested into Google Earth Engine. Nodata value was coded as 0 the same as the authors.

Citation

Chen, X., Wang, Z., Zhang, F., Shen, G., & Chen, Q. (2024). A global annual simulated VIIRS nighttime light dataset from 1992 to 2023. Scientific Data, 11(1380).
https://doi.org/10.1038/s41597-024-04228-6

Dataset Citation

Chen, Xiuxiu; Zhang, Feng; Wang, Zeyu (2023). A history reconstructed time series (1992-2011) of annual global NPP-VIIRS-like nighttime light data through a super-resolution
U-Net model. figshare. Dataset. https://doi.org/10.6084/m9.figshare.22262545.v8

srunet_npp_viirs_ntl

Earth Engine Snippet

var viirs_ntl = ee.ImageCollection("projects/sat-io/open-datasets/srunet-npp-viirs-ntl");

Sample code: https://code.earthengine.google.com/?scriptPath=users/sat-io/awesome-gee-catalog-examples:global-utilities-assets-amenities/GLOBAL-SRUNET-NPP-VIIRS-LIKE-NTL

License

This dataset is made available under Creative Commons 4.0 Internal Attribution License.

Created by: Chen, Xiuxiu et al.

Curated in GEE by: Samapriya Roy

Keywords: Nighttime light, VIIRS, NTL, NPP-VIIRS, Urban monitoring, super resolution, machine learning

Last updated: 2024-01-28

Logo Ask AI