In the context of intensifying climate change, it is particularly important to assess the transformation of spring floods as a key phase of the hydrological regime of rivers. This study provides a comprehensive analysis of the characteristics of maximum runoff in the Zhaiyk–Caspian basin for the modern period and projected changes for 2030, 2040, and 2050 based on CMIP6 climate scenarios (SSP3-7.0 and SSP5-8.5). Analysis of observations at 34 hydrological stations showed a reduction in spring runoff by up to 35%, a decrease in the duration of high water and a reduction in maximum water discharge on some rivers by up to 45%. It has been established that those rising temperatures, more frequent thaws, and reduced autumn moisture lead to lower maximum water discharge and a redistribution of the seasonal flow regime. Scenario projections revealed significant spatial heterogeneity: some rivers are expected to experience an increase in maximum discharge of up to 72%, while others will see a steady decline in maximum discharge of up to 35%. The results obtained indicate the need to transition to an adaptive water management system focused on the regional characteristics of river basins and the sensitivity of small- and medium-sized watercourses to climate change.

The possibility of group simulation of monthly runoff hydrograph by canonical expansion method is demonstrated. The simulation procedure takes into account the coefficients of autocorrelation and mutual correlation and enables one to reproduce the ratios between parameters C _ν and C _s for monthly time intervals. The model preserves the similarity between within-year distributions of the measured and simulated series.

Studied is the dependence of the entropy on the series of annual runoff on the number of observation years. Plotted are nomograms for estimating the unbiased value of the entropy depending on the number of observation years and number of gradations. Nomograms are plotted from the simulated series for the Pearson type III distribution.

The article considers the possibility of using the results of joint modeling of runoff and runoff- forming factors to assess and forecast the water resources in the Arys River basin. To solve this problem, the canonical decomposition method was used, which allows us to characterize a random process (river runoff) by a set of independent random variables and non–random functions. The method is optimal in the class of linear transformations, decomposition was carried out according to natural orthogonal components. Monthly, annual and vegetative hydrograph of the Arys River runoff were predicted for 2030, 2040 and 2050, according to two scenarios, RCP 4.5 and RCP 8.5. The results of the study showed that changes in river runoff by 2030, 2040, and 2050 fluctuate around the climatic norm of 1974–2019, both downwards and upwards. The use of the canonical decomposition model based on the dependence of runoff on climatic characteristics is fully justified by the results obtained.