Long‐Term Water Imbalances of Watersheds Resulting From Biases in Hydroclimatic Data Sets for Water Budget Analyses

Assessing the water budget closures and source of water budget imbalances is fundamental to improving the understanding of changes in the hydrological system and their associated impacts. We analyzed the long‐term (1982–2016) water budget for 1,561 watersheds by using various observed data sets for precipitation (P), evapotranspiration (ET), observed streamflow (Q), and total water storage change (∆TWS). The results show that 93%, 79%, 44%, and 20% of watersheds show water imbalances ratio less than 30%, 20%, 10%, and 5% of their corresponding precipitation. The average absolute water imbalance ratio for all watersheds is 14.2% of P. Watersheds showing large water imbalance ratio values are mostly located in biomes of Tropical and Subtropical Moist Broadleaf Forests, Boreal Forests/Taiga, and Tundra. Different P, ET, and Q data set combinations result in different degrees of water imbalance. The water budget imbalance ratio shows a significant negative relationship with humidity index and vegetation coverage, while a positive relationship with the proportions of irrigation area and watershed area. Showing small water imbalances for most watersheds, reanalysis precipitation data set (ERA5 and MSWEP), MTE evapotranspiration data set performed better than other data sets in water budget analyses in most biomes. The uncertainties of P, ET, Q, and ∆TWSGRACE contribute to 40.1%, 14.3%, 26.6%, and 19% of the water budget imbalance on average, respectively. Improving the accuracy of P and ET estimates, and streamflow measurements are critical to better understanding the water budget and improves modeling of hydrological processes. Key Points Water budget imbalances are less than 30%, 20%, 10%, and 5% of precipitation for 93%, 79%, 44%, and 20% of all watersheds Reanalysis precipitation (ERA5 and MSWEP) and MTE evapotranspiration perform better than other data sets in the water budget in most biomes Moist Broadleaf Forests, Boreal Forests/Taiga, and Tundra show the largest water imbalance ratio due to bias in hydroclimatic data sets