INTRODUCTION

Nowadays, one of the most urgent problems faced by researchers of Arctic and subarctic territories is how permafrost thawing can affect the release of organic matter from the soils and peats. Organic matter coming to the hydrosphere as a consequence of the thawing of permafrost rocks affects the geochemistry of the aquatic systems (Moiseenko et al., 2017; Pokrovsky et al., 2020). Studies have shown that the contribution of terrestrial organic carbon to freshwater water bodies significantly increases relative to the biota of these water bodies themselves due to permafrost degradation (Wauthy et al., 2018). Isotope data also indicate the boost of the mobilization of the carbon buried in permafrost rocks, which enriches the waters of Arctic rivers (Feng et al., 2013). In addition to the effect on the carbon balance itself, processes initiated by global climatic changes facilitate the release of many other elements from permafrost rocks (Stepanova et al., 2015; Pokrovsky et al., 2020; Moiseenko et al., 2020; Lim et al., 2021, 2022). In addition, it remains uncertain what is the scale of the transfer of organic matter and chemical elements from soils and peats into aquatic systems and the forms of their migration. This problem provokes keen interest not only in the context of understanding how the chemical composition of surface continental aquatic systems is formed (Savichev et al., 2011; Savichev and Moiseeva, 2016; Savichev et al., 2016; Krickov et al., 2020; Manasypov et al., 2020; Moiseenko et al., 2020; Ivanova et al., 2021) but also concerning the analysis of material transfer from continental aquatic systems to shelf areas (Holmes et al., 2012; Perminova et al., 2019; Shiklomanov et al., 2021). The authors of this publication studied the chemical composition of surface waters (rivers, thermokarst lakes, and thaw depressions) in the Yamalo–Nenets Autonomous District (YaNAD) in the Arctic zone of Western Siberia. These studies were focused primarily on the behavior of dissolved organic carbon (DOC) and biogenic elements. Moreover, studying the chemical composition of surface waters in the area is not only of academic but also of applied significance. Surface water in the area are utilized for a public water supply. According to (State Report…, 2021), water in 14 of 25 surface water objects (i.e., more than 50%) used for public water supply in the YaNAD territory in 2020 failed to meet sanitary and hygienic standards.

The study of factors that can control the chemical composition of natural waters and the estimation of their current ecological−geochemical state is thus an important aspect of the evaluation of the anthropogenic impacts on the water bodies and the further prediction of the possible transformations of the chemical composition of the waters under the effect of climatic changes, which are the most intense in northern territories (Reshet’ko and Moiseeva, 2016; Shiklomanov et al., 2021).

STUDY AREA

The study area is located in the Yamalo–Nenets Autonomous District and is constrained within the Arctic zone of the Russian Federation (Presidential Decree…, 2014). Waters were sampled in the catchment areas of large rivers: Ob, Taz, Pur, and Nadym.

The plain surface topography of the area and the submeridional trends of the orographic features of most of the northern Western Siberia, in combination with the low elevations (35–55 m), hamper the fast runoff of the ice-melting and rain waters (Syso, 2007; Voskresenskii, 1962) and are thus favorable for the bogging of the territory (Karavaeva, 1982).

Geologically, the area is constrained within the young epi-Hercynian Western Siberian plate. The lower structural−tectonic floor comprises basement rocks, and the upper one consists of the Meso-Cenozoic sedimentary cover sequences. Most of the study area is covered by Late and Middle Quaternary marine, glacial, and glacial−marine plains made up mostly of middle-loamy sediments (Lazukov, 1970), which give way to Middle and Late Quaternary sandy and sandy−loamy sediments of lacustrine−alluvial and alluvial genesis (Atlas…, 1971). The rocks contain such minerals as chlorite and hydromicas, for example, glauconite, montmorillonite-group minerals, iron hydroxides, pyrite, and siderite.

The soil cover is dominated by peaty, mostly oligotrophic, and more rare eutrophic soils with thick peat horizons (Shamilishvili et al., 2016). The temperature regime of peat horizons in the soils, which compose the top parts of the soil profile, efficiently controls the depths at which the permafrost occurs. The thawing permafrost rocks become part of the soil profiles, whereas the elements released are involved in soil and landscape geochemical processes. The thickness of the seasonal-thaw layer is 50–85 cm, because of the geographic position of the territory, its surface topography, and the composition of its plant communities.

METHODS OF THE FIELDWORK, CHEMICAL ANALYSIS, AND STATISTICAL PROCESSING OF THE RESULTS

The fieldwork was carried out in September 2020, in the YaNAD territory. During the work, four study areas of key importance were organized. During the fieldwork, 47 bottom water samples were taken, including 23 samples from water courses (rivers and streams) and 24 from areal water bodies (lakes and thaw depressions). Herein we understand a thaw depression as a microtopographic feature formed when ice-rich permafrost rocks thaw, and the resultant topographic features are filled with water. This process describes the initial evolution of thermokarst lakes (Grosse et al., 2013).

The study objects were selected to most comprehensively study the lake−stream−river systems in the catchment areas of major YaNAD rivers and to analyze the chemical composition of waters in the water objects of various order. Figure 1 shows the distribution of the sampling sites of the surface waters. At each sampling site, we measured rapidly varying parameters of the waters: temperature, pH, electric conductivity (EC), and Eh (HI98194 and HANNA Instruments). Water samples for chemical analysis for main components were stored in 1-L plastic bottles, which had been precleaned triply with the water to be analyzed. Water samples to be analyzed for trace elements and dissolved organic carbon (DOC) were filtered through membrane acetate-cellulose filters with 0.45 μm pores into falcon polypropylene tubes (45 mL) and glass vessels (100 mL), respectively. The water samples were stored in 500-mL glass vessels and conserved with chloroform to study nitrogen and phosphorus compounds. Water samples for analysis of the chemical oxygen demand (COD) and the permanganate index (PI) were collected in 200-mL glass vessels and acidified with sulfuric acid. Before their delivery to the laboratory, the samples were stored in a shadowed place at a temperature close to 4oC.

Fig. 1.
figure 1

Sampling sites of surface waters in the territory of YaNAD in 2020. Sampled catchment areas: (1) Taz River; (2) Pur River; (3) Ob River; (4) Nadym River.

Full size image

The chemical composition of the natural water samples was analyzed at the Laboratory of Hydrogeochemistry of the Tomsk Polytechnic University, Tomsk, Russia. Concentrations of CO2, \({\text{CO}}_{3}^{{2 - }}\), \({\text{HCO}}_{3}^{ - }\), and Ca2+ were determined by titration. Concentrations of the ammonium ion in natural waters were determined photometrically on a KFK photocolorimeter (ZOMZ, Russia). Concentrations of Na+, K+, \({\text{NO}}_{3}^{ - }\), \({\text{NO}}_{2}^{ - }\), \({\text{PO}}_{4}^{{3 - }}\), \({\text{SO}}_{4}^{{2 - }}\), and Cl− were determined by ion chromatography (on an ICS-5000, Dionex, USA). Concentrations of Mg2+ were calculated from the total hardness and Ca2+ concentration. The value COD was determined photometrically (on a Fluorat 02-3M, Lumex, Russia). The PI value was evaluated by titration. The concentration of DOC was analyzed by high-temperature catalytic oxidation on a Vario TOC cube (Elementar, Germany).

The raw data were processed by mathematical statistics using the MS Excel 2016 and Statistica 10 software. The chemical types of the waters were identified regarding components with concentrations of ≥25 mg-equiv. %, which are listed in descending order of their concentrations. The background concentrations were calculated following the distribution law. For statistical analysis, values defined as below the detection limit were substituted for values equal to the halves of the detection limits of the components. The distribution types (normal or lognormal) of the datasets were identified based on the following criteria: the visual correspondence of the histogram of the dataset to a certain distribution type (1), the visual correspondence of the quantile plot of the dataset to a certain distribution type (2), and by conducting a chi-square (3), Lillieforce (modified Kolmogorov–Smirnov) (4), and Shapiro–Wilk tests (5). If more than half of these criteria confirmed the hypothesis that the distribution is normal, then the mathematical expectation (background value) was assumed to be the arithmetic mean. The geometric mean was assumed for estimating the background values for components whose behavior was described by a lognormal distribution. If the distribution law had not been reliably identified, then the background concentration was defined based on an expert analysis, which involved a visual estimation of the histogram of the dataset, the mean-square deviation, standard errors and outliers in boxplots, and the number of values below the detection limit. The background concentrations for datasets whose distribution laws were unknown was assumed to be the geometric mean if the distribution of the random value was close to a lognormal one. Otherwise, the mathematical expectation was assumed as the median value. If a dataset included relatively many values below the detection limits, which precluded a reliable identification of the distribution law, then the background values were assumed as those lower than the detection limits.

We carried out correlation analysis by the Spearman rank correlation coefficient to reveal relationships between components of the water chemical composition because the distributions of most of the physicochemical parameters were different from normal, and the datasets were not large. The mean values of several independent datasets of physicochemical parameters of various objects (rivers, streams, lakes, and thaw depressions) were compared using a nonparametric analog of single-factor analysis of variance: the Kruskal–Wallis test with the study object as the grouping variable. We applied the Mann–Whitney U test, a nonparametric alternative to the t test for two independent datasets, to compare the means of two independent data groups. The structure of interrelations between the variables (physicochemical parameters of water composition) and observations (study objects) was determined and unobvious common factors that explain these interrelations were identified by classifying the variables by principal component analysis (PCA), using a factorial plane.

RESULTS AND DISCUSSION

Chemical Composition of Surface Waters in the Arctic Zone of Western Siberia

The studied waters show broad variations in their pH, from weakly acidic to weakly alkaline, at an average pH 5.78 (Table 1). The waters are mostly weakly acidic and near-neutral. The waters are ultrafresh, and their mineralization (total dissolved solids, TDS) exceeds 100 mg/L only at two sites: in the Malaya Khadyr’yakha and Ob rivers.

Table 1. Principal physicochemical parameters of the chemical composition of surface waters sampled in YaNAD in September 2020
Full size table

The waters show relatively high parameters reflecting concentrations of organic compounds. For example, the chemical oxygen demand (COD) values reach 414 mgО2/dm3, and this parameter at the great majority of the sampling sites was higher than the standardized maximum permissible concentrations for surface waters for recreational use, 15 mgО2/dm3 (SanPiN 1.2.3685-21). Samples from the site with the highest COD also show the highest permanganate index (PI) of 168 mgО2/dm3 (thaw depression 1020, Table 1). The maximum dissolved organic carbon (DOC) concentration (95.2 mg/L) is also typical of a thaw depression (1015). Most of the highest DOC concentrations are typical of thaw depressions and, more rarely, lakes and streams. A similar dependence was also found for COD and PI. Our data on DOC concentrations are generally consistent with values previously found in rivers and thermokarst lakes in zones of continuous and discontinuous permafrost rocks in the Western Siberian Lowland (Manasypov et al., 2015, 2020; Pokrovsky et al., 2015). CO2 concentrations, which provide circumstantial evidence that the organic matter is mineralized, in the waters lie within the range of 1.9–32.6 mg/L at an average of 9.7 mg/L. These values correspond to CO2 concentrations determined in (Manasypov et al., 2020) in spring- and summertime in the waters of thermokarst lakes in the continuous permafrost zone and are much higher than the values found in the waters of thermokarst lakes in the discontinuous permafrost zone.

The cationic composition of the waters is variable, with the most widely spread waters being calcic–magnesian and magnesian–calcic, and sodium only occasionally found as the dominant cation. It should be mentioned that the waters contain relatively high ammonium concentrations. The N–\({\text{N}}{{{\text{H}}}_{4}}^{ + }\) concentrations vary from 0.05 to 3.03 mg/L at an average of 0.3 mg/L. Concentrations of N–\({\text{N}}{{{\text{H}}}_{4}}^{ + }\) at 19 sampling sites exceeded the standardized maximum permissible concentrations for water objects for fish farming, 0.4 mg/L (Order…, 2016). The maximum concentrations of the ammonium ion were found in thaw depressions 1015 and 1020 (Table 1). Along with the thaw depressions, high ammonium concentrations are also typical of the lakes and are occasionally found in large rivers, such as Pur and Pankit’yakha. Waters from these water objects show high COD values (>80 mgО2/dm3), indicating that the natural waters become enriched in the ammonium ion when the organic matter is mineralized. Because of the low TDS values of the waters, the fraction of the ammonium ion locally reaches 70% of the cations. This indicates that organic matter plays an important role in controlling the composition of the waters because the ammonium ion is the first step in the mineralization of nitrogen-bearing organic matter (Li et al., 2012; Zhu et al., 2015; Sorokin and Aleksandrov, 2013). This statement is confirmed by the fact that ammonium is dominant in the nitrogen balance in the waters, which indicates that organic matter is continuously brought to the waters and that the geochemical conditions are favorable for its accumulation (absence of dissolved oxygen and low rate of water exchange). It should also be mentioned that N–\({\text{N}}{{{\text{H}}}_{4}}^{ + }\) content in the surface water is much higher than the concentration values obtained on samples taken in 2015 by (Vorobyev et al., 2017).

The anions are usually dominated by the hydrocarbonate ion: some of the waters are hydrocarbonate, and others (rarer) are hydrocarbonate–sulfate and hydrocarbonate–chloride. Anions in samples from some sites are dominated by the sulfate ion: some waters are sulfate, while others (rarer) are sulfate–hydrocarbonate and sulfate–chloride. Anions in samples from two sit es were dominated by the chloride ion (a thaw depression and Kirill-Vislor Lake). Elevated chloride-ion concentrations are constrained within areas with abundant oil- and gas-producing facilities and the influence zones of oil and gas pipelines, and hence, they are most probably anthropogenically induced. This conclusion is based on analysis of long-term observations in oil-producing areas (Moskovchenko, 2007; Moskovchenko et al., 2008). Elevated (relative to the background) chloride-ion concentrations were primarily found in thaw depressions and rivers. The concentrations of the nitrite ion (recalculated to nitrogen) are no higher than 0.006 mg/L, and the concentrations of \({\text{N - NO}}_{3}^{ - }\) are only rarely higher than 0.05 mg/L. The highest concentrations of nitrates were found in the river waters. The phosphate ion concentrations are also relatively low and only occasionally exceed 0.05 mg/L in recalculation to phosphorus (maximum permissible concentration for oligotrophic water objects for recreational use) (Order…, 2016). These samples were taken mostly from rivers in the basin of the Pur River near the urban-type settlement of Urengoi. The engineering development of this territory most probably predetermines this feature of the chemical composition of the waters. High phosphate concentrations are typical of the fourth marine terrace, whose rocks are utilized in the infrastructure of the hydrocarbon fields (landfills at well clusters and access roads) and can be brought to water objects with the surface and groundwater runoff. The concentrations of nitrates and phosphates are generally consistent with data from (Vorobyev et al., 2017) on small rivers in the basins of the Ob, Taz, and Pur rivers that flow through zones with discontinuous permafrost.

Results of Statistical Processing

To test the statistical significance of the aforementioned relations and trends and to identify more complicated relations and factors that can influence the chemical composition of the waters, we applied several statistical techniques.

Correlation analysis revealed positive correlations between the dominant cations and anions with pH (Table 2). The only exception was the sulfate ion, which shows no correlation with pH. These correlations reflect processes of interaction between water and rock (mineral substratum). Thereby the absence of correlation between the concentration of the sulfate ion and pH provides evidence of processes that affect \({\text{S}}{{{\text{O}}}_{4}}{{^{2}}^{ - }}\) transfer into the water and are more complicated than simple dissolution of the rock. The concentrations of dominant cations and anions are also correlated with different strengths. The strongest correlations are typical of the Ca2+ and Mg2+ concentrations. The weakest correlations occur between K+ and \({\text{S}}{{{\text{O}}}_{4}}{{^{2}}^{ - }}\), Cl–. Systematically occurring correlations were found of the CO2 concentration with pH and with the concentrations of such main ions, which is explained by the balance of dissolved species (CO2, \({\text{HC}}{{{\text{O}}}_{{\text{3}}}}^{ - }\), and \({\text{C}}{{{\text{O}}}_{{\text{3}}}}{{^{{\text{2}}}}^{{\text{ - }}}}\)) depending on the pH of the solution. The strong correlation between the concentrations of the chloride and sodium ions is worth mentioning. This correlation is more robust than that between Cl– and all other cations, anions, and TDS, indicating that the waters may enrich in Na+ due to anthropogenic impact, along with the Cl– ion (Moskovchenko, 2007).

Table 2.   Spearman rank correlation coefficients between physicochemical parameters of the chemical composition of surface waters in YaNAD
Full size table

Parameters reflecting the influx of organic matter (DOC, COD, and PI) strongly correlate with one another (r = 0.93 between DOC and COD, r = 0.88 between COD and PI, and r = 0.87 between DOC and PI) and show a structure of correlation practically identical to that with other components and compositional parameters of the surface waters. A strong positive correlation was found for DOC, COD, and PI with the concentration of ammonium, whose correlation structure is practically identical to that of PI (Table 2). Negative correlations of DOC, COD, and PI were found with the concentrations of the dominant cations, \({\text{HCO}}_{3}^{ - }\), pH, and TDS, and a weak negative correlation is also typical of nitrate. The nitrate concentrations also show a weak positive correlation with the concentrations of K+, \({\text{HCO}}_{3}^{ - }\), pH, and TDS.

With regard toConcerning the identified relations, we have estimated the differences between the groups of the studied water objects. In this instance, the major groups were those of water courses (rivers and streams), water bodies (lakes), and thaw depressions (which are mainly spot objects). The Kruskal–Wallis test for comparing several independent datasets has shown that differences between the three groups are statistically significant in values of pH, \({\text{HCO}}_{3}^{ - }\), \({\text{S}}{{{\text{O}}}_{4}}{{^{2}}^{ - }}\), Ca2+, Mg2+, Na+, \({\text{N - NH}}_{4}^{ + }\), DOC, COD, PI, and TDS (Figs. 2, 3). Then it was necessary to test whether these differences occur between the three groups or only in certain pairs. To do this, we used the Mann–Whitney U test to compare two independent samples. We have determined that all of the aforementioned differences are statistically significant between water courses and thaw depressions, except for the sulfate ion concentration. Thereby the averages and dispersions of the compositional parameters (pH, \({\text{HCO}}_{3}^{ - }\), \({\text{S}}{{{\text{O}}}_{4}}{{^{2}}^{ - }}\), Ca2+, Mg2+, Na+, and TDS) in the water courses are higher than the values of the waters of the thaw depressions (Fig. 2), whereas the averages and dispersions of \({\text{N - NH}}_{4}^{ + }\), DOC, COD, and PI are higher for the waters of the thaw depressions (Fig. 3).

Fig. 2.
figure 2

Differences between principal physicochemical parameters of the chemical composition of water bodies in YaNAD.

Full size image
Fig. 3.
figure 3

Differences between parameters reflecting nitrogen and carbon concentrations in water bodies in YaNAD. See Fig. 2 for symbol explanations.

Full size image

The only statistically significant differences between the water courses and lakes are the differences between \({\text{HCO}}_{3}^{ - }\), \({\text{S}}{{{\text{O}}}_{4}}{{^{2}}^{ - }}\), Ca2+, Mg2+, Na+, pH, and TDS. The average dispersions of all the aforementioned parameters for the water courses are higher than for lakes (Fig. 2). No statistically significant differences between lakes and thaw depressions were found in \({\text{HCO}}_{3}^{ - }\), \({\text{N - NH}}_{4}^{ + }\), DOC, pH, COD, and PI. The averages and dispersions of pH and \({\text{HCO}}_{3}^{ - }\) concentrations are higher for lakes, whereas the waters of the thaw depressions are characterized by notably higher averages and dispersions for \({\text{N - NH}}_{4}^{ + }\), DOC, COD, and PI than the waters of lakes.

It is thus possible to identify the principal distinguishing features of the water objects from one another (Fig. 4). The water courses (rivers and streams) differ from lakes and thaw depressions in wider ranges of pH values (from acidic to weakly alkaline) and elevated TDS. Additionally, the thaw depressions differ from all other water objects in having more acidic pH and elevated parameters related to the presence of organic matter: COD, PI, and concentrations of \({\text{N - NH}}_{4}^{ + }\) and DOC, with these features predetermined by the specifics of the origin of these water objects.

Fig. 4.
figure 4

Structure of the classification characteristics of water objects in YaNAD.

Full size image

We have also analyzed the probability of significant differences between the compositions of surface waters in the various river basins. The water objects studied under this research project belong to the following large river basins: the catchment areas of the Ob, Taz, Pur, and Nadym rivers. Using the Kruskal–Wallis test, we have determined that no statistically significant correlations occur between differences in the physicochemical parameters of waters in these four basins, except only Ca concentrations.

According to the Kaiser test (Kaiser, 1960), the chemical composition of surface waters in YaNAD is controlled mostly by three factors: the eigenvalues of the dispersion of the three factors are 5.9, 2.4, and 1.0, respectively. As variables for analyses (with basic observations) with the PCA method, we selected the following compositional parameters and components for which statistically significant differences were determined between the water objects of different types: pH, \({\text{HCO}}_{3}^{ - }\), \({\text{S}}{{{\text{O}}}_{4}}{{^{2}}^{ - }}\), Ca2+, Mg2+, Na+, \({\text{N - NH}}_{4}^{ + }\), DOC, COD, PI, and TDS. All other components were assumed as additional variables. The analysis was based on the data of all observations.

The three-factor model is able to explain 85.2% variations in the original data. The first factor positively correlates with variables that reflect variations in the general chemical composition of the waters (pH, \({\text{HCO}}_{3}^{ - }\), Ca2+, Mg2+, Na+, K+, TDS, and not so much Cl–) and shows a weaker negative correlation with \({\text{N - NH}}_{4}^{ + }\), DOC, and COD (Fig. 5, Table 3). This factor accounts for 53.6% of the explained dispersion.

Fig. 5.
figure 5

Projections of the variables on the factorial plane (Factor 1 vs. Factor 2).

Full size image
Table 3.   Factor loadings
Full size table

The maximum loadings of the second factor fall onto parameters that reflect contents of organic matter (\({\text{N - NH}}_{4}^{ + }\), COD, and PI) and more weakly correlate with the DOC concentration itself. The second factor explains 22.1% of the dispersion of the data. The third factor explains only 9.5% variations in the data and is related to the \({\text{S}}{{{\text{O}}}_{4}}{{^{2}}^{ - }}\) concentration through a strong negative correlation (Table 3).

Projecting factor loadings on the factorial plane of the first and second factors, we have identified all three clearly discernible types of the water objects: water courses, lakes, and thaw depressions (Fig. 6). It is also evident that the first factor explains mostly the dispersion of data on the linear water objects (rivers and springs), whereas the second factor explains the dispersion of data on the thaw depressions. This distribution of the factor loadings confirms the conclusion about the causes of the factors. For example, the conclusion is confirmed that the second factor is related to the local thawing of organogenic soil horizons. As follows from the distribution of the loadings, the third factor most appreciably affects the water courses and not so much the thaw depressions, the water objects do not group as obviously as for the first and second factors.

Fig. 6.
figure 6

Distribution of the factor loadings of Factor 1 and Factor 2 with regard to the type of the water objects.

Full size image

It is reasonable to conclude that the first factor reflects the effects of interaction between the water and mineral constituents of the underlying rocks and, for the water courses, also likely the distance for which the suspended material is transported. Also, the first factor somehow reflects to a certain extent the composition of the atmospheric precipitation, which is one of the principally important water sources for the water courses and large surface water basins. In this context, it should be mentioned that the water was collected during the low-water period in summer–autumn when the role of atmospheric precipitation as a source of water was relatively small. The second factor is related to the thawing of organogenic (peat) soil horizons. The most probable reasons for the individualization of the third factor are compositional features of the rocks with which the waters interact, as well as processes related to the activities of microorganisms. The effects of the anthropogenic factor can be neglected in this situation because of the absence of correlations between the concentrations of the sulfate and chloride ions and because the correlation with the sodium ion is of the same level as those of the other main cations and anions, which are of natural origin. Available data can hardly be sufficient for a more detailed and accurate description of the third factor that controls the chemical composition of the waters.

CONCLUSIONS

Ultrafresh surface waters are widespread in the Yamalo–Nenets Autonomous District, and their pH is commonly neutral or weakly acidic. Hydrocarbonate ion dominates the anions of the waters, and calcium and magnesium dominate the cations. The waters also contain relatively high ammonium concentrations, which strongly positively correlate with DOC, COD, and PI.

Water courses (rivers and streams) in YaNAD differ from the water bodies (lakes and thaw depressions) in having higher pH values and concentrations of main components (\({\text{HCO}}_{3}^{ - }\), Ca2+, Mg2+, and Na+), which are caused mostly by natural factors, namely, by interaction between the waters with underlying rocks and soils. Among the water bodies, thaw depressions differ from lakes in having higher DOC concentrations and values of the other parameters related to the inflow of organic compounds (\({\text{N - NH}}_{4}^{ + }\),COD, and PI), which indicates that the thawing of organogenic soil horizons significantly affects the chemical composition of the waters.

Statistical analysis has shown that the principal factors controlling the surface water composition are the interaction of water with the mineral components of the underlying rocks and organic matter, which is brought from organogenic soil horizons. The former factor most strongly affects the streams, and the second one controls mostly the composition of the thaw depressions. The increase in the concentrations of the chloride and sodium ions at some of the sampling sites is explained by an anthropogenic impact. The processes and factors controlling the supply of the sulfate ion to the waters can hardly be inferred from currently available data. However, the most probable reasons for the elevated sulfate concentrations are some compositional features of the rocks drained by the streams and processes related to the activity of microorganisms.