A Synthesis of Global Streamflow characteristics, Hydrometeorology, and catchment Attributes (GSHA) for Large Sample River-Centric Studies

Abstract. Our understanding and predictive capability of streamflow processes largely rely on high-quality datasets that depict a river’s upstream basin characteristics. Recent proliferation of large sample hydrology (LSH) datasets has promoted model parameter estimation and data-driven analyses of the hydrological processes worldwide, yet existing LSH is still insufficient in terms of sample coverage, uncertainty estimates, and dynamic descriptions of anthropogenic activities. To bridge the gap, we contribute the Synthesis of Global Streamflow characteristics, Hydrometeorology, and catchment Attributes (GSHA) to complement existing LSH datasets, which covers 21,568 watersheds from 13 agencies for as long as 43 years based on discharge observations scraped from web. In addition to annual streamflow indices, each basin’s daily meteorological variables (i.e., precipitation, 2 m air temperature, longwave/shortwave radiation, wind speed, actual and potential evapotranspiration), daily-weekly water storage terms (i.e., snow water equivalence, soil moisture, groundwater percentage), and yearly dynamic descriptors of the land surface characteristics (i.e., urban/cropland/forest fractions, leaf area index, reservoir storage and degree of regulation) are also provided by combining openly available remote sensing and reanalysis datasets. The uncertainties of all meteorological variables are estimated with independent data sources. Our analyses revealed the following insights: (i) the meteorological data uncertainties vary across variables and geographical regions, and the prominent patterns revealed should be accounted for by LSH users, (ii) ~6 % watersheds shifted between human managed and natural states during the GSHA time span, which may be useful for hydrologic analysis that takes the changing land surface characteristics into account, and (iii) GSHA watersheds observed a more widespread declining trend in runoff coefficient than an increasing trend, which warrants further studies on water availability. Overall, GSHA is expected to serve hydrological model parameter estimation and data-driven analyses as it continues to improve. GSHA v1.0 can be accessed at https://doi.org/10.5281/zenodo.8090704 (Yin et al., 2023).