Year-to-year and many-year runoff variations in world rivers

Data on the annual, maximal, and minimal runoff for about 3000 gauges, taken from international archives, as well as data on total runoff into the World Ocean are given. The rivers for which the gauges considered in the study are outlet gauges were divided into 6 groups by their nourishment type, lake abundance, and the degree of potential anthropogenic impact on runoff regime. Twenty-two quantitative input river basin quantitative parameters and runoff characteristics, as well as 102 basic numerical characteristics evaluated by the author, including parameters of stochastic models, identified for each series, were analyzed for each runoff series. To study the effect of the non-Gaussian distribution of runoff series on analysis results, calculations for each series were carried out twice: for original series and for series converted into those with normal distribution. It is shown that most series of annual and maximal runoff in lake-free rivers whose regime shows no anthropogenic disturbance (including rivers with glacier nourishment) can be adequately described, at a time scale of the order of decades, by a time series of mutually uncorrelated random variables. Models of first-order autoregression process and higher order models are predominant in the description of annual runoff series of lake-type rivers and rivers, whose basins contain large reservoirs and irrigation systems, as well as in the description of minimal-runoff series of all types of rivers.