Low water balance consistency of state-of-the-art hydrological datasets

Hydrological research benefits from a growing number and diversity of hydrological datasets. At the same time, the consistency across the increasing suite of datasets is unclear, limiting the comparability of findings derived with different datasets and variables. Here, we find overall low consistency of numerous state-of-the-art precipitation, evapotranspiration, runoff, and soil moisture datasets in terms of the water balance. Consistency is inferred between variations in soil moisture and in precipitation minus evapotranspiration minus runoff, where datasets are combined with independent datasets representing the remaining water balance variables. Highest consistency in the case of precipitation datasets is generally found for satellite-based datasets, while gauge-based datasets performed better in Northern Hemisphere regions with dense in-situ observations. In the case of evapotranspiration, highest consistency is found for satellite-based and reanalysis datasets, and in the case of runoff for gauge-based and reanalysis datasets. Reanalysis soil moisture datasets that consider deep soil water dynamics show higher consistency than satellite-based or gauge-based datasets. Spatial variations of consistency are mostly related to aridity and temperature as they influence precipitation measurement quality. Soil moisture dataset consistency is additionally affected by vegetation cover. We find widespread increases in dataset consistency in the northern mid-latitudes during the study period, probably related to climate warming.