Streamflow predictions are vital for detecting flood and drought events. Such predictions are even more critical to Sub-Saharan African regions that are vulnerable to the increasing frequency and intensity of such events. These regions are sparsely gaged, with few available gaging stations that are often plagued with missing data due to various causes, such as harsh environmental conditions and constrained operational resources. This work presents a novel workflow for predicting streamflow in the presence of missing gage observations. We leverage bias correction of the Group on Earth Observations Global Water and Sustainability Initiative ECMWF streamflow service (GESS) forecasts for missing data imputation and predict future streamflow using the state-of-the-art temporal fusion transformers (TFTs) at 10 river gaging stations in the Benin Republic. We show by simulating missingness in a testing period that GESS forecasts have a significant bias that results in poor imputation performance over the 10 Beninese stations. Our findings suggest that overall bias correction by Elastic Net and Gaussian Process regression achieves superior performance relative to traditional imputation by established methods. We also show that the TFT yields high predictive skill and further provides explanations for predictions through the weights of its attention mechanism. The findings of this work provide a basis for integrating Global streamflow prediction model data and the state-of-the-art machine learning models into operational early-warning decision-making systems in resource-constrained countries vulnerable to drought and flooding due to extreme weather events.

Holocene streamflow was reconstructed for two rivers on Vancouver Island, British Columbia, Canada in 500-yr intervals. The San Juan River watershed is located on the wetter western side of the island, whereas the Koksilah River watershed is positioned on the drier eastern side. Both watersheds are forested. To reconstruct streamflow, temporal changes in precipitation (estimated using a pollen-based transfer function) and evapotranspiration were established for each watershed and integrated into a water balance model, calibrated using modern data. While seasonal streamflow variability was maintained throughout the Holocene, with greater flow in the winter relative to the summer, the amount of discharge has changed markedly through time. Lowest simulated flow occurred in the earliest Holocene, with low-flow conditions beginning earlier in the year and extending later into the fall. Such conditions may have inhibited salmon from using many of the smaller rivers in the region. Streamflow steadily increased throughout the early Holocene so that by ca. 6500 cal yr before present near-modern flow regimes were established. As climate changes in the future, the San Juan and Koksilah watersheds are expected to remain as pluvial hydroclimatic regimes, though with an extended season of low flow similar to conditions during the early Holocene.

Gauged river flow records from China generally span only a few decades, which hampers the detection of long-term, decadal- to centennial-scale cycles and trends in streamflow variability. New and updated tree-ring chronologies help reconstructed the water-year (October–September) streamflow for the Aksu River, which is an important river at the edge of the Taklimakan Desert that drains into the Tarim Basin. The reconstruction dates back to 1692 and has an adjusted r2 of 0.61 (1957–2006). Based on frequency, intensity and duration of droughts and pluvial events, the lowest streamflows occurred in the 1920s. Since then streamflow has continuously increased, and was exceptionally rapidly after the 1960s, until today. The start and end of the 20th century to the present were the highest streamflow periods. The mid-20th century was the longest and driest period over the past 300 yr. The reconstructed streamflow series has a strong positive correlation with the North Atlantic Oscillation Index. Changes in mid-latitude circulation patterns influencing precipitation may have indirectly resulted in streamflow variations along the Aksu River over the past 300 yr. The rapid increase and the exceptional streamflows of the 1960s are likely linked with global warming and mid-latitude atmospheric circulation changes.

Non-negative time series which are first-order autoregressive with a mixed exponential innovation process are studied. Properties and approximate marginal distributions for such series are found. Modifications to include exact zeroes and to increase variability so that the time series is a more realistic model of rivers which are dry for part of the year are discussed.