A global hyperspectral soil albedo dataset for land surface modelling: Global Hyperspectral Soil Albedo Dataset (GHSAD) - SoilGrids 2.0

The Global Hyperspectral Soil Albedo Dataset (GHSAD) is aimed at developing a global hyperspectral soil albedo dataset applicable to land surface models through machine learning methods. We propose an integrated framework for hyperspectral soil albedo modeling, and this framework utilizes machine learning methods to construct the global hyperspectral soil albedo dataset. This dataset is created based on the SoilGrids 2.0 soil properties dataset. The dataset produced based on GSDE is available at: 10.5281/zenodo.18909426.

GHSAD provides global hyperspectral albedos of dry and wet soils, with a spatial resolution of 0.05°×0.05°, a spectral resolution of 10 nanometers, and a range from 400 to 2500 nanometers. GHSAD offers soil albedos under different soil moisture conditions. We discretize the range of volumetric water content from dry soil to saturated soil into ten equal intervals, and then calculate the soil albedo for each interval.

There are two ways for users to utilize this dataset: (1) interpolate soil albedo values at different soil moisture levels; (2) calculate the actual soil albedo based on the dry soil albedo provided by GHSAD in combination with the soil water balance model (BSM) or other more complex wet soil albedo schemes. This dataset is stored in the Network Common Data Format (netCDF).