To seek stochastic analogies in key processes related to the hydrological cycle, an extended collection of several billions of data values from hundred thousands of worldwide stations is used in this work. The examined processes are the near-surface hourly temperature, dew point, relative humidity, sea level pressure, and atmospheric wind speed, as well as the hourly/daily streamflow and precipitation. Through the use of robust stochastic metrics such as the K-moments and a second-order climacogram (i.e., variance of the averaged process vs. scale), it is found that several stochastic similarities exist in both the marginal structure, in terms of the first four moments, and in the second-order dependence structure. Stochastic similarities are also detected among the examined processes, forming a specific hierarchy among their marginal and dependence structures, similar to the one in the hydrological cycle. Finally, similarities are also traced to the isotropic and nearly Gaussian turbulence, as analyzed through extensive lab recordings of grid turbulence and of turbulent buoyant jet along the axis, which resembles the turbulent shear and buoyant regime that dominates and drives the hydrological-cycle processes in the boundary layer. The results are found to be consistent with other studies in literature such as solar radiation, ocean waves, and evaporation, and they can be also justified by the principle of maximum entropy. Therefore, they allow for the development of a universal stochastic view of the hydrological-cycle under the Hurst–Kolmogorov dynamics, with marginal structures extending from nearly Gaussian to Pareto-type tail behavior, and with dependence structures exhibiting roughness (fractal) behavior at small scales, long-term persistence at large scales, and a transient behavior at intermediate scales.

Machine learning has been employed successfully as a tool virtually in every scientific and technological field. In hydrology, machine learning models first appeared as simple feed-forward networks that were used for short-term forecasting, and have evolved into complex models that can take into account even the static features of catchments, imitating the hydrological experience. Recent studies have found machine learning models to be robust and efficient, frequently outperforming the standard hydrological models (both conceptual and physically based). However, and despite some recent efforts, the results of the machine learning models require significant effort to interpret and derive inferences. Furthermore, all successful applications of machine learning in hydrology are based on networks of fairly complex topology that require significant computational power and CPU time to train. For these reasons, the value of the standard hydrological models remains indisputable. In this study, we suggest employing machine learning models not as a substitute for hydrological models, but as an independent tool to assess their performance. We argue that this approach can help to unveil the anomalies in catchment data that do not fit in the employed hydrological model structure or configuration, and to deal with them without compromising the understanding of the underlying physical processes.

Comprehending the correlation between alterations in human‐flood distance and flood fatalities (as well as displacements) is pivotal for formulating effective human adaptive strategies in response to floods. However, this relationship remains inadequately explored in existing global analyses. To address this gap, we examine 910 flood events occurring from 2000 to 2018, resulting in significant numbers of fatalities and displacements. We find that in 53% of countries, humans tend to distance from floods, particularly in the Middle East. Such distancing greatly mitigates flood fatalities and displacements. Simultaneously, in areas with increased flood protection level (FPL), humans are less likely to move away from floods. Furthermore, FPL and human‐flood distance have decreased in regions affected by ice jam‐ and hurricane‐induced floods from 2000 to 2018. Notably, regions with human‐flood distance slightly below the average for a given flood type experience more severe flood fatalities.