Search

While global streamflow reanalysis has been evaluated at different spatial scales to facilitate practical applications, its local performance in the time–frequency domain is yet to be investigated. This paper presents a novel decomposition approach to evaluating streamflow reanalysis by combining wavelet transform with machine learning. Specifically, the time series of streamflow reanalysis and observation are respectively decomposed and then the approximation components of reanalysis are evaluated against those of observed streamflow. Furthermore, the accumulated local effects are derived to showcase the influences of catchment attributes on the performance of streamflow reanalysis at different scales. For streamflow reanalysis generated by the Global Flood Awareness System, a case study is devised based on streamflow observations from the Catchment Attributes and Meteorology for Large-sample Studies. The results highlight that the reanalysis tends to be more effective in characterizing seasonal, annual and multi-annual features than daily, weekly and monthly features. The Kling–Gupta efficiency (KGE) values of original time series and approximation components are primarily influenced by precipitation seasonality. High values of KGE tend to be observed in catchments where there is more precipitation in winter, which can be due to low evaporation that results in reasonable simulations of soil moisture and baseflow processes. The longitude, mean precipitation and mean slope also influence the local performance of approximation components. On the other hand, attributes on geology, soils and vegetation appear to play a relatively small part in the performance of approximation components. Overall, this paper provides useful information for practical applications of global streamflow reanalysis. [ABSTRACT FROM AUTHOR]

Predicting flood event classes aids in the comprehensive investigation of flood behavior dynamics and supports flood early warning and emergency plan development. Existing studies have mainly focused on historical flood event classification and the prediction of flood hydrographs or certain metrics (e.g., magnitude and timing) but have not focused on predicting flood event classes. Our study proposes a new approach for predicting flood event classes based on the class membership functions of flood regime metrics and hydrological modeling. The approach is validated using 1446 unimpacted flood events in 68 headstream catchments widely distributed across China. The new approach performs well, with class hit rates of 68.3% ± 0.4% for all events; 65.8% ± 0.6%, 56.8% ± 0.9%, and 69.5% ± 0.9% for the small, moderate and high spike flood event classes, respectively; and 82.5% ± 1.2% and 75.4% ± 1.1% for the moderate and high dumpy flood event classes, respectively. Furthermore, it performs better in the basins of northern China than in those of southern China, particularly for the small spike flood event class in the Songliao and Yellow River Basins, with hit rates of 80.0% ± 3.2% and 78.8% ± 3.2%, respectively. Our results indicate that the new approach will help improve the prediction performance of flood events and their corresponding classes, and provide deep insights into the comprehensive dynamic patterns of flood events for early warning and control management. Plain Language Summary: The prediction of flood event classes is more effective and informative than that of individual events for obtaining comprehensive dynamic characteristics of flood events for early warning and development of emergency plans. However, this is challenging owing to massive flood events with remarkable spatiotemporal heterogeneity and unclear class membership functions. We propose a class prediction approach for flood events using the class membership functions of flood behavior metrics based on frequency analysis and catchment hydrological modeling. This approach deeply mines the behavior characteristics of historical flood events with a strong mathematical basis and cooperates with hydrological models to predict flood classes. Over thousand unimpacted flood events in 68 headstream catchments across China were selected to validate the robustness of the prediction scheme. Our results show that the proposed approach well predicts the flood class with class hit rates of 68.3% ± 0.4% for all events; 65.8% ± 0.6%, 56.8% ± 0.9%, and 69.5% ± 0.9% for the small, moderate and high spike flood event classes, respectively; and 82.5% ± 1.2% and 75.4% ± 1.1% for the moderate and high dumpy flood event classes, respectively. Our study has strong implications for scientific flood warning research and flood mitigation management. Key Points: A prediction approach for flood event classes is developed using class membership functions of flood regime metrics with hydrological modelClass membership degree is estimated using the frequency distribution function, which fits well with statistical significanceAverage class hit rates are 68.3% ± 0.4% for all the events and predictions of dumpy flood class are better than those of spike flood classes [ABSTRACT FROM AUTHOR]