Figure captions
Figure 1. Map of the upper Terek basin with topography, glacierized areas, cropland, and river network. The main 33 gauging stations considered in this study are represented with dots. See Table 1 for label transcriptions. The extent of the glaciers is shown as of 2011-2016 (Raup et al. , 2007), cropland as of 2015 (Buchneret al. , 2020)
Figure 2. Mean monthly values of suspended sediment discharge (SSD , kg·s−1) measured at the gauging stations. Colored shaded areas represent the range corresponding to ± standard error. Mean values and standard errors are computed over the entire observation period. The length of the period is shown in brackets.
Figure 3 . Suspended sediment load change time series. The black line displays the time series with the uncertainty envelope. The red line corresponds to a linear trend.
Figure 4. CUSUM charts for the mean annual SSD . The transition years are indicated with grey (Taylor, 2000) and red (Pettitt, 1979) lines depending on the changepoint test.
Figure 5. Double mass curves of precipitation–sediment during 1958–2018 in the upper Terek basin. See Table 4 for changepoint location (i.e., transition year). The Red dashed line shows an extrapolated cumulative sediment discharge.
Figure 6. Density plots of trend magnitudes for SSL (% yr−1). Here we plotted randomly simulated SSL trend values for the whole observation period (see Section 2.6 ). The vertical dashed lines indicate trends with the highest kernel density, and a black vertical line highlights the zero trend.
Figure 7. Scatter plots of the suspended sediment load trends (% change per year) versus the catchment area (a), glacier area in 1986 (b), cropland, and forest area in 1987 (c, d). Circles indicate the mean simulated SSL trend values, while error bars indicate the 95% confidence interval (see Section 2.6).
Figure 8. Spearman’s rank correlation coefficients (Spearman r) between the estimated SSL trend values (Eq.(3)) and the considered potential controlling factors for different altitude zones. Each boxplot shows the distribution of 1000 Spearman r-values, where each value was obtained by randomly simulating a set of SSL trend values and calculating the correlation between these values and the indicated variable (see Section 2.6). Thus, whiskers of each boxplot represent 1.5 times the difference between the 75% and 25% quantile. Altitude is the actual height of the gauging station above the mean sea level; Area is a catchment area; Cropland and forest represented their fraction in a catchment in 1987; Glacier stood for glacier fraction in a catchment in 1986.