There is a drastic geographic imbalance in available global streamflow gauge and catchment property data, with additional large variations in data characteristics. As a result, models calibrated in one region cannot normally be migrated to another without significant modifications. Currently in these regions, non‐transferable machine learning models are habitually trained over small local data sets. Here we show that transfer learning (TL), in the senses of weight initialization and weight freezing, allows long short‐term memory (LSTM) streamflow models that were pretrained over the conterminous United States (CONUS, the source data set) to be transferred to catchments on other continents (the target regions), without the need for extensive catchment attributes available at the target location. We demonstrate this possibility for regions where data are dense (664 basins in Great Britain), moderately dense (49 basins in central Chile), and scarce with only remotely sensed attributes available (5 basins in China). In both China and Chile, the TL models showed significantly elevated performance compared to locally trained models using all basins. The benefits of TL increased with the amount of available data in the source data set, and seemed to be more pronounced with greater physiographic diversity. The benefits from TL were greater than from pretraining LSTM using the outputs from an uncalibrated hydrologic model. These results suggest hydrologic data around the world have commonalities which could be leveraged by deep learning, and synergies can be had with a simple modification of the current workflows, greatly expanding the reach of existing big data. Finally, this work diversified existing global streamflow benchmarks.

The impacts of reservoirs, especially multiple reservoirs, on the flow regimes and ecosystems of rivers have received increasing attention. The most widely used metrics to quantify the characteristics of flow regime alterations are the indicators of hydrologic alteration (IHAs) which include 33 parameters. Due to the difference in the degree of alteration and the intercorrelation among IHA parameters, the conventional method of evaluating IHA parameters that assigns the same weight to each indicator is obviously inadequate. A revised IHA method is proposed by utilizing the projection pursuit (PP) and real-coded accelerated genetic algorithm (RAGA). Data reliability is analyzed by using the length of record (LOR) method. The projection values reflecting the comprehensive characteristics of the evaluation parameters are calculated. Based on these methods, a scientific and reliable evaluation of the cumulative impacts of cascading reservoirs on the flow regime was made by examining the Jinsha River. The results showed that with the increase in the number of reservoirs, the alteration degrees of IHA parameters gradually increased in groups 1, 2, 3 and 4 but decreased in group 5 (each group addresses the magnitude, timing, frequency, duration and rate of change in turn), and the flow duration curves showed a declining trend at the high-flow part and an increasing trend at the low-flow part. The flow regime alteration of the outlet section was more stable than before. This change had a negative impact on downstream fish reproduction and ecological protection. An attempt at ecological regulation was made to simulate the natural rising process of water, and four major Chinese carps have a positive response to the flood peak process caused by manual regulation.