Dobrovolsky S.G. Global Changes in River Runoff. М.: GEOS, 2011. – 660 p. ISBN 978-5-89118-558-6
The aim of the work was an analysis of the long-term, year-to-year and many year-changes in the annual, maximal and minimal runoff all over the World. Four spatial scales were investigated: local (at about 3000 river gauges), within 23 main rivers of the World, at the scale of continents and macro-regions (such as the territory of the Russian Federation, foreign Europe and foreign Asia), and finally at global scale. A new classification of watersheds was proposed with the respect to the types of feeding, and the calculations were executed separately for each type. A new system of estimating statistical and stochastic parameters was proposed on the basis of the random functions theory and on a new algorithm of generating the Gaussian random values. Within the framework of this approach, new formulas for calculating the dispersion (standard deviations), autocorrelations, the orders of stochastic (autoregressive models) were proposed. A new two-sided criteria for verification the zero-hypothesis on the stationarity/nonstationarity of the river runoff time series (separately for the mathematical expectation, standard deviations, and autodorrelations) were introduced as well. The results of the work show that, unlike previous researches, the Markov first-order model does not play the dominant role in describing long-term changes in the annual river runoff. Instead, it was shown, that for the rivers of the main type of feeding (not lake-rivers and not rivers regulated by large dams) the zero-order autoregressive model (“white noise”) plays the most important role. However, for the rivers regulated by lakes and by artificial dams the “simple Marov chain” is still preferable. The analysis of the degree of stationarity demonstrated that for the rivers of the main type only 6% of the annual runoff time series demonstrate the nonstationary dynamics.
660 pages, 210 references, 474 graphs including 158 maps, 37 tables.