The accurate estimation of sediment flux in naturally confluent rivers remains challenging. In this study, we developed a momentum‐sediment model (MSM) that considers the complex structure of confluent rivers to estimate sediment flux. Trend analyses of annual runoff and sediment flux from 1956 to 2022 were conducted at four hydrological stations in the upper reaches of the Yangtze River Basin: Pingshan, Zhutuo, Cuntan, and Yichang. Runoff change was insignificant, whereas the sediment flux decreased significantly at all four stations. The entire time series was divided into five periods: a reference period (P1: 1956–1984) and four changing periods (P2: 1985–1993; P3: 1994–2002; P4: 2003–2012; and P5: 2013–2022). The hydrological time series of the reference period served as the calibration period for constructing the momentum‐sediment models for Zhutuo, Cuntan, and Yichang stations. The MSM expressions were validated in the four changing periods. The momentum‐sediment models for period P1 at Zhutuo and Cuntan stations could be directly applied in periods P2–P5. The MSM for the Yichang station was improved by incorporating the trapping efficiency of the Three Gorges Reservoir. The simulation results of the MSM for sediment flux outperformed the sediment rating curves and linear relationship, particularly for maximum measured values. The applicability of the MSM was validated in the Jialing River, which is characterized by a small basin area, and the Yellow River, which is known for its scarce water and rich sediments. This confirmed that the MSM was insensitive to watershed scale and hydrological conditions.

As a widespread natural disaster, drought has a significant impact on changes in regional vegetation cover. Based on 1‐month standardized precipitation index (SPI1), this study took the Yellow River basin (YRB) with frequent droughts as the research object, using NDVI data from 1999 to 2018, Mann–Kendall trend test, and correlation test to study the temporal and spatial characteristics of meteorological drought, explored the response of vegetation cover to meteorological drought, and further discussed the possible factors affecting meteorological drought. The results of the study showed that from 1999 to 2018, the SPI1 in the YRB had not increased significantly in a large area in different months (p > .05). On the monthly time scale, the NDVI of the YRB gradually increased from north to south, showing a significant increasing trend the central area (p < .05). And in the summer months, there was a significant positive correlation between SPI1 and NDVI in some areas (p < .05). In the meantime, sunspots, ENSO, and PDO had significant effects on the SPI1 in the YRB (p < 0.05). Furthermore, in areas with frequent extreme droughts, the average value of NDVI was high, which was mainly because the area was in a cultivated vegetation area with a wide‐scale policy of returning farmland to forest and afforestation activities. Returning farmland to forests and afforestation activities will help overcome the impact of meteorological drought and other adverse conditions on vegetation. Therefore, returning farmland to forests can be implemented for ecological restoration in arid areas in the YRB.

Potential evapotranspiration (PET) is an important input variable of many ecohydrological models, but commonly used empirical models usually input numerous meteorological factors. In consideration of machine learning for complex nonlinear learning, we evaluated the applicability of three machine learning algorithms in PET estimation in the Yellow River basin (YRB), in addition to determining significant factors affecting the accuracy of machine learning. Furthermore, the importance of meteorological factors at varying altitudes and drought index grades for PET simulation were evaluated. The results show that the accuracy of PET simulation in the YRB depends on the input of various meteorological factors; however, machine learning models including average temperature (T_mean) and sunshine hours (n) as input achieved satisfactory accuracy in the absence of complete meteorological data. Random forest generally performed best among all investigated models, followed by extreme learning machine, whereas empirical models overestimated or underestimated PET. The importance index shows that T_mean is the most influential factor with respect to PET, followed by n, and the influence of T_mean on PET gradually decreased with increased altitude and drier climate, whereas the influence of n shows the opposite trend.