Djankuat Glacier Station in the North Caucasus, Russia: A Database of complex glaciological, hydrological, meteorological observations and stable isotopes sampling results during 2007-2017

The study presents a dataset on long-term complex glaciological, hydrological, meteorological observations and isotopes sampling in an extremely underreported alpine zone of the North Caucasus. The Djankuat research basin is of 9.1 km2, situated on elevations between 2500 – 4000 m, by 30% covered with glaciers. The biggest in the basin – the Djankuat 20 glacier was chosen as representative of the central North Caucasus during the International Hydrological Decade and is one of 30 ‘reference’ glaciers in the world that have annual mass balance series longer than 50 years (Zemp et al., 2009). The dataset covers 2007–2017 and contains the result of yearly measurements of snow thickness and density; dynamics of snow and ice melting; measurements of water runoff, conductivity, turbidity, temperature, δ18O, δ2H on the main gauging station (844 samples in sum) with a one-hour or several-hours step depending on the parameter; data on δ18O and δ2H sampling of liquid 25 precipitation, snow, ice, firn, groundwater in different parts of the watershed regularly in time during melting season (485 samples in sum); precipitation amount, air temperature, relative humidity, shortwave incoming and reflected radiation, longwave downward and upward radiation, atmospheric pressure, wind speed and direction – measured on several automatic weather stations within the basin with 15 min – one-hour step; gradient meteorological measurements to estimate turbulent fluxes of heat and moisture, measuring three components of wind speed at a frequency of 10 hertz to estimate the turbulent 30 impulse heat fluxes over the glacier surface by the eddy covariance method. All the observations were done during ablation period (June–September) and were interrupted in winter. The dataset was published on knb.ecoinformatics.org long-term repository doi:10.5063/F1H1307Q and will be further updated. The dataset can be useful for developing and verifying hydrological, glaciological and meteorological models for high elevation territories, to study impact of climate change on Earth Syst. Sci. Data Discuss., https://doi.org/10.5194/essd-2018-124 O pe n A cc es s Earth System Science Data D icu ssio n s Manuscript under review for journal Earth Syst. Sci. Data Discussion started: 9 October 2018 c © Author(s) 2018. CC BY 4.0 License.

to study the impact of climate change on hydrology of mountain regions, using isotopic and hydrochemical approaches to study mountain hydrology.As the dataset includes the measurements of hydrometeorological and glaciological parameters during the catastrophic proglacial lake outburst in the neighboring Bashkara valley in September 2017, it is a valuable contribution to the study of this dangerous hydrological phenomenon.

Introduction
The important role of mountains territories and their high sensitivity to climate change is concluded in vast amount of recent research (Dyurgerov, 2003, Weingartner et al., 2007, Auer et al., 2007, Viviroli et al., 2011, Pachauri et al., 2014, Zemp et al., 2015).However, it is widely recognized that there is still a great lack of observational data on climate, glaciers and hydrology of mountain areas (Gietl, 1990, Barry, 1992, Singh et al., 1999, Global change…, 2001, Schaefli et al., 2005, Bales et al., 2006,).The density of hydrological stations in the world's mountainous regions is from 3 (in Europe) to 100 (in Asia) times lower than those recommended by the World Meteorological Organization (Viviroli et al., 2011, Bobrovitskaya, Kokorev, 2014,).The majority of field observations in mountainous catchments are conducted in Scandinavia, the Alps and the mountains of the USA, while vast Asian territories and the Southern Hemisphere stay extremely understudied (Barry, 1992, Dyurgerov, 2003, Meier et al., 2003, Zemp et al., 2009, Viviroli et al., 2011, Immerzeel et al., 2012,).The Great Caucasus, that used to have a developed observational network during the Soviet Union period, recently joined the above-mentioned poorly studied territories in terms of level of information availability on meteorological and glacio-hydrological topics (Barry, 1992, Dyurgerov, 2003, Shahgedanova et al., 2005, Bobrovitskaya, Kokorev, 2014).
The Djankuat research basin, 9.1 km 2 in area, is located at 43.2⁰N and 42.75⁰E in the alpine zone of the North Caucasus (Russia), between 2600 and 4000 m (Fig. 1).Djankuat glacier, occupying 27% of its area, was chosen as representative of the central North Caucasus during the International Hydrological Decade (Boyarsky, 1978).The mass-balance measurements have been carried out on Djankuat glacier since 1967 till now without interruption (www.wgms.ch).Glaciological observations are carried out by standard methods (Østrem andBrugman, 1992, Boyarsky, 1978).Hydrological and meteorological complex measurements were included in the monitoring program of the station during the International Hydrological Decade and were terminated in the end of 1970 s (Boyarsky, 1978).The complex hydrometeorological observation were resumed in the Djankuat research basin under the initiative of the collective of the authors since 2007.The covered time period by hydrometeorological measurements during the ablation season of each year and the observational program gradually increased during [2007][2008][2009][2010][2011][2012][2013][2014][2015][2016][2017] and now goes beyond the standard network hydrological and meteorological observations.The relative cutting of the program in 2011-2012 was related to a special military regime that was applied in Kabardino-Balkaria republic by the Government.
There are 4 main locations in the basin where the Automatic Weather Stations (AWS) are being installed (see Fig. 1, Table 1).
All the meteorological stations operate only during the ablation season of each year.Two Campbell AWS are located in the central part of the Djankuat glacier above the ice surface (AWS1) and the debris of the glacier (AWS2).AWS1 operated through the period 2007-2017 (excluding 2011).The second Campbell station, located over the debris (AWS2) was operating for three years (2007)(2008)(2009).The both stations provide measurements of the air temperature, relative humidity, downward and upward shortwave radiation, downward and upward longwave radiation, wind speed, wind direction, atmospheric pressure.A Davis weather stations operated at Base Camp (Base Camp AWS) through 2007-2009 and 2013-2017  The dataset was published on a long-term data repository (doi:10.1594/PANGAEA.894807) and will be updated as the observations in the basin are still ongoing (as of autumn 2018).Some of the data, presented in the study was already successfully used in scientific work (Zemp et al., 2009, Rets, Kireeva, 2010, Lambrecht et al., 2011, Zemp et al., 2011, Zemp et al., 2015, Popovnin, Pylayeva, 2015, Rets et al., 2017, Toropov et al., 2017, Chernomorets et al., 2018)

Study Area
The Greater Caucasus stretches 1300 km along the border between Russia and Georgia from the Black Sea to the Caspian Sea.
The alpine zone extends above the orographic snowline which height is approximately at 2000 m, the highest point of the Greater Caucasus is Elbrus mountain (5642 m).The climate here is moderate continental to high-alpine.The main centers of atmospheric influence for the North Caucasus are extensions of the Icelandic depression from the west and the Siberian high from the east during the winter period and extensions of the Azores during the summer (Volodicheva, 2002, Shagedanova et al., 2005).Influence of the Black Sea is superimposed on the general circulation.Together with the complex orographic effects they results in complex spatial precipitation distribution and strong precipitation gradients.The precipitation decreases both southeastwards and with a decrease in elevation.Annual precipitation sum varies from 200-400 mm in Eastern plain part and 600-800 mm in Western plain part to 800-1300 and more in mountainous part (Rets at al, 2018).
Great spatial variety is characteristic of river runoff in the North Caucasus.The water supply of the region is strongly dominated by runoff formed in the high mountains, a unit of area at an elevation of 3-4 km can be 10 times more productive in terms of water yield than the lowlands.In the most alpine zone of the North Caucasus annual unit discharge varies from 20-30 to 50-60 liters/(s*sq.km).In the foothills mean annual runoff unit discharge sharply declines to 5-15 liters/(s*sq.km).The vast plain territories do not add much to the total runoff of rivers: the values of unit discharge decrease gradually in the Northeast direction down to zero or become negative (Rets et al., 2018).
Rivers with a substantial share of the alpine zone in the total area of the watershed are characterized by a high-water period lasting from late spring to September, and stable winter low-flow period.With a decrease in elevation the share of snowmelt in river runoff diminishes, the beginning of high-water and winter low-flow periods shifts to earlier dates, rain floods start playing a more substantial role in maximum discharges and a winter low flow period is more often interrupted by snowmelt winter floods.Annual water regime of rivers in the lowland territory of the North Caucasus depends on annual distribution of precipitation.In the central and Eastern North Caucasian plain territory precipitation occurs mostly in summer that results in summer high-flow period and both winter and summer low-flow periods.Winter precipitation maximum is characteristic of the Western part of North Caucasus (Rets et al., 2018).
The Djankuat research basin is situated on the northern slope of the central part of the Main Caucasian Ridge (see Fig. 1).It is a typical alpine watershed of 9 km 2 with the elevation range 2600 -4000 m, with steep slopes (more than 20⁰ in average) and nival-glacial landscape (Fig. 2).An overall exposition of the basin is the North-North-West.In the years 2007-2017 glaciers occupied 30% of the territory of the basin.The main glacier with the same name -Djankuat glacier -is the source of the Djankuat river.It is a valley glacier, with the lowest point of the tongue at approximately 2750 m, the elevation of the bergschrund is at 3600 m.The mean elevation of the glacier is 3210 m, the area is 2.6 km 2 and its length is 3.0 km.The maximum measured thickness of the glacier is 105 m, and the average thickness is 31 m (Lavrentiev et al., 2014).The Djankuat river basin also contains three small glaciers with areas less than 0.5 km 2 : Koyavgan, Via-Tau, and Visyachiy.These glaciers contribute runoff to the Djankuat river upstream the main gauging station (see Fig. 1).The Djankuat River is a source of the Adul-Su River -a tributary of the Baksan River which drains into the Caspian Sea via the Terek river.
The uppermost gauging station on the Baksan River was situated 12 km from its source in Usengi.The river basin area at the gauging station is 180 km 2 .The mean annual discharge of the Baksan River at Usengi gauging station is 9.9 m 3 /sec that amounts to mean unit annual discharge of 55 liters per sec.per km 2 (Rets, Kireeva, 2010).The water-abundant period of the Basksan River in the upstream is prolonged and steady, it extends from May to September-October.The general wave of runoff hydrograph, formed by snow and ice melting, is overlain with sharp peaks of rain floods.The maximum water levels are usually recorded in July.A stable winter low-flow period with minimum monthly discharge of 2.4 m 3 /sec is observed in February to March (Rets, Kireeva, 2010).
The climate of the Djankuat research basin is characterized by a distinct seasonality in temperature.The mean monthly air temperatures at the Terskol, the closest to the research basin all-year meteorological station, situated 16 km northwest of the study area at an altitude of 2146 m goes below zero during November-December.The warmest months are July and August with mean monthly temperatures above 12 ⁰C (Fig. 3).Monthly precipitation sums are by 40-50% higher in the warm period of the year (May-September) than during winter (Fig. 3).The annual precipitation at the Terskol weather station varied from 590 to 1330 mm with a mean annual value of 950 mm.Daily precipitation maxima occur in July to September in response to convective activity triggered by a combination of strong insolation and depressions developing on the Polar front and enhanced by the orographic uplift (Shagedanova, 2002).
The observations in the Djankuat research basin, included in the presented dataset, were carried out under the conditions of slightly warmer summer periods in comparison to the long-term average, and substantially higher amounts of precipitation, especially during the spring period -from March to May (Fig. 3).The strong influence of spring snowfalls during the observation period also is reflected in the results of Djankuat glacier snow sampling for stable isotopes (Rets et al., 2017).The outlined tendency is likely to be the consequence of climate change in the region.
According to the majority of studies (Alekseev et al., 2014, Toropov et al., 2018a, Rets, Kireeva, 2010) a statistically significant positive trend in air temperature amounting to 0.7 -1 ⁰С/10 year is observed during the summer period in the North Caucasus.
According to Rets and Kireeva (2010) this tendency is more evident in the plain territory and foothills.A slight positive trend in the mean temperature in the ablation period (May to September), 0.3 ⁰C/10 year, is observed at the Terskol meteorological station since the end of 1970 th (Fig. 4).The timing of alteration in the air temperature tendencies regime corresponds well with the situation observed in the European territory of Russia, where he time period starting from 1978 is identified as a "contemporary period" in term of the recent climate forced changes in river runoff regime (Frolova et al., 2014, Rets et al., 2018).
In the winter period the observed tendencies in air temperature in the North Caucasus are very inhomogeneous: Alekseev et al. ( 2014) report a statistically insignificant positive trend.Toropov et al. (2018a) claim a statistically significant rise in air temperature of the winter period which is observed in the Eastern Caucasus, close to the Caspian Sea.In the study by Rets and Kireeva (2010) a decrease in air temperature of the winter period was revealed in the mountainous part of the North Caucasus.
According to the on the Terskol meteorological station the value of mean air temperature during the accumulation season (October to April) remains stable in the study area (Fig. 4).
According to different studies, a positive trend in annual precipitation sum of 5%/10 year is reported by Alekseev et al. (2014), and no statically significant trend is reported for the most of the North Caucasus by Toropov et al. (2018a).An increase in annual precipitation sum was revealed for the majority of mountainous stations and a number of foothills located in the central part of Northern Caucasus (Rets, Kireeva, 2010).At the Terskol meteorological station the amount of precipitation is constantly rising during the whole observational period (Fig. 5).The increase in annual sum is 3.5%/10 years is different for the ablation (2.1%/10 year) and accumulation (5%/10 year) period.The most intensive rise in precipitation is observed in spring (8.6%/10 year in March, 7%/10 year in April) and autumn (10.3%/10 year in October).This result is consistent with the result reported by Alekseev et al. (2014) for the whole territory of the North Caucasus.
The intensive degradation of glaciation is observed in the North Caucasus (Zemp et al., 2015, Shahgedanova et al., 2014).The area of glaciers in the North Caucasus dropped by 12.6% during 1970-2000(Voitkovskiy et al., 2004), and by 4.7% between 2000and 2010/2012(Shahgedanova et al., 2014), amounting to approximately 17% in total during 1970-2012.The glaciers terminus retreat increased from the 1987-2000/2001 period to the 2000/2001-2010 period by the factor 2.5-3.8.The highest recession rates of 11-14 m yr −1 were observed in the central Main Caucasus ridge and on Mountain Elbrus.The largest total retreat was registered for the Bolshoi Azau glacier, located on Mt.Elbrus.This glacier lost 500 m during 1987-2010, retreating at a steady rate of 22 m yr −1 (Shahgedanova et al., 2014).Glacier retreat and the increase in supraglacial debris cover is also accompanied by the emergence and growth of proglacial lakes and related increase in proglacial lake outburst floods (Stokes et al., 2007).On the 1 st of September 2017 an outburst of Bashkara lake in the upstream of neighboring to the Djankuat basin valley gave a rise to catastrophic mudflow that led to major destructions and human casualties (Chernomorets et al.,2018).
Annual river runoff in the mountainous part of the North Caucasus shows a slight positive trend during 1940-2010.In the most elevated areas the long-term mean value of annual runoff remains stable.This is in contrast to the plain part and the foothills in the North Caucasus, where the annual runoff increased by 30-70% during last 3 decades (Rets et al., 2018).An increase in amount, duration and extent of thaws and general reduction of the annual cold period duration in the lowest elevation belts of the North Caucasus is reflected in a 50-100% rise in minimum monthly discharges in winter.In mountainous areas long-term oscillation of winter minimum monthly discharge strongly depends on local factors, such as geological structure.In the upper reaches of some tributaries of the Terek and Kuban River positive trends are still not observed, while in neighboring valleys long-term variations of winter minimum monthly discharges correlate with the corresponding variations in the foothills and on plain.On the highest elevation belts, where the temperature is still strongly negative in winter, winter minimum monthly discharge remains stable on the long-term scale (Rets et al., 2018).

Discharge measurements
Water discharge of the Djankuat River at the gauging station (see Fig. 1, see Table 1) is calculated at an hourly time step from the water level using a rating curve Q=f(H) (Table 2).At the Djankuat gauging station water level is measured with 10 min to 1-hour time step (depending on the year of observation) by means of an automatic water level logger with a pressure sensor (ADU-02 during 2007-2013, Solinst Level logger in 2014-2017).The level logger is placed with a ripple shield in an artificial bay constructed at the river bank (Fig. 6).Control water level measurements are made 6-7 times a day by a staff gauge.Rating curves are redrawn for each month of each ablation season (Fig. 7).A dilution method with NaCl as a tracer was used for discharge measurements as turbulent flow conditions make it impossible to apply the current meter (Dobriyal et al., 2017).
As there is a possibility of erroneous results due to the loss or incomplete mixing of the tracer arising from the difference in velocity in the upper and lower layers of the stream (Dobriyal et al., 2017), every discharge measurement is repeated twice.
The value is accepted, if the difference between the two simultaneous measurements does not exceed 10%.For the most of the discharge measurements at the Djankuat gauging station this difference was less than 5%.50-80 water discharges are measured every ablation season to draw a rating curve.
Water discharge at the Djankuat gauging station mostly stayed in the range of 1 to 2 m 3 /s during the 2007-2017 observational period (Fig. 8a), the mean value of water discharge was 1.39 m 3 /s (Table 3).Low frequency water discharges (less than 1% of duration) lay in the range of 3.5 to 8.46 m 3 /s.The maximum discharge, 8.46 m 3 /s, was observed on the 1 st of July 2015 at 9:00.
It was a result of a strong rain flood caused by 227 mm of precipitation in sum over 7 days superimposed on an intensive snow and ice melting in the river basin.
The inter-annual fluctuations of the Djankuat River runoff can be quite substantial (Fig. 9).The Djankuat River was the most abundant in water during 2015-2016, the least in 2013.Mean water discharge for June to September in 2015 (1.88 m 3 /s) was twice as large as in 2013 (0.97 m 3 /s).The Djankuat River is most abundant in water in July (Fig. 9).Mean monthly discharge in this month is 1.3-2.8m 3 /s.June and August are comparable in terms of mean discharge, 0.93 to 1.8 and 1.0 to 2.0 m 3 /s, respectively.In September runoff due to ice and firn melting decreases with the decrease in the incoming solar radiation, the seasonal melt water is gradually drained from the Djankuat River basin.Mean monthly discharge is 0.6-1 m 3 /s.At the end of September, the ablation period ends with the first stable fresh snow cover on the glacier.
The Djankuat River hydrograph has a saw-tooth shape with a pronounced daily maximum and minimum typical for glacial rivers (Fig. 10).A diurnal fluctuation of discharge is great and can be compared with the overall seasonal fluctuation: up to 1.5-2 m 3 /s on a day without rain.The rise of a rain flash-flood can be very intensive: more than 1 m 3 /s in an hour.

Electrical Conductivity and Salinity
Electrical conductivity (Cond) of water was measured at the Djankuat gauging station 6-7 times a day in 2014 and 2017 with an Electrical conductivity meter (conductometer Econics Expert-002).When a conductometer with a logger function was used in 2015-2016 the measurement was done with a 1-hour time step (Table 2).Water salinity is calculated from the conductivity measurements, using a dependency Salinity=f(Cond).The dependency was drawn in 2013 using the data on simultaneous measurement of electrical conductivity and complete chemical analysis in 19 samples with conductivity from 4.2 to 87.5 µS/cm (Fig. 11).
Total amount of 3464 electrical conductivity measurements was done at the Djankuat gauging station during 2007-2017 (Table 3).The Djankuat River water is low-mineralized, the value of electrical conductivity stayed in the range of 55-85 µS/cm for 90% of the time (Fig. 9d).The electrical conductivity value strongly depends on the percentage of snow and ice melt water in the total river runoff.During long periods without rain with intensive melting the water of the Djankuat River can be diluted up to 40-50 µS/cm during day time.On the daily minimums of water discharge in early morning, the electrical conductivity rises by 10-30 µS/cm.At the end of the ablation season, when melting is strongly reduced, the electrical conductivity of the Djankuat River reaches values of 110 to 115 µS/cm during night-morning hours, which is close to the value of electrical conductivity of groundwater in the basin.

Water Temperature
Water temperature was measured at the Djankuat gauging station 6-7 times a day in 2017 with a water temperature sensor built in a conductometer Econics Expert-002.When a conductometer with a logger function was used in 2015 and 2016 the measurement was done with a 1-hour time step (Table 2).Total amount of 3259 measurements was made (Table 3).Water

Water Turbidity
Turbidity of the Djankuat River was measured at the gauging station 6-7 times a day during 2015 to 2017.The regularity of measurement is defined for each month according to the shape of a diurnal hydrograph.At the event of heavy rainfall, the measurements are performed every 15 minutes, and averaged in the database of 1-hour time step.Some first test measurements of turbidity were made in 2008 and 2013.Optical turbidity was measured by a portable turbidimeter Hach 2100P in Nephelometric Turbidity Units (NTU).The values of turbidity in weight units (g/m 3 ) were calculated using a dependency Weight Turbidity=f(Optical Turbidity).The dependency was drawn in 2015 and 2016 using the data on simultaneous measurement of optical turbidity and weight turbidity analysis in 19 samples with optical turbidity from 66.7 to 36400 NTU.
The total amount of 1991 measurements is included in the database (Table 3).The Djankuat River turbidity has an extremely uneven distribution (Fig 9b): staying less than 400-500 NTU (250-350 g/m 3 ) most of the time, on the event of a heavy rainfall (more than 20 mm/day) water turbidity can abruptly rise to 1000-5000 NTU (750-4000 g/m 3 ) and even 30 000 -40 000 NTU (25 000-33 000 g/m 3 ) for several hours (Fig. 10).These values are in the same range as the values of water turbidity registered in such river as Huanghe (Zhang, Huang, 1993).The maximum value of turbidity (45 060 NTU or 37 200 g/m 3 ) was measured on 1 st of September 2017 after 87 mm of rain with average intensity of 30 mm/hour.The same rain event triggered of an outburst of Bashkara lake in the upstream of a neighboring valley, that gave a rise to a catastrophic mudflow (Chernomorets et al., 2018).

Stable isotopes
The first test sampling of stable isotopes content in the Djankuat River was carried out in 2013 (Table 2).Its main goal was to define a needed regularity of sampling to get a representative mean daily value of δ 18 O and δD.As the daily variation of δ 18 O and δD turned out to be low compared to other hydrological parameters of the Djankuat River, sampling was done twice a day in 2014-2017: on the maximum and minimum of water level.The total amount of 844 samples was taken from the Djankuat River at the gauging station during this period.
A clear-cut difference in isotopic composition of ice/snow meltwater and liquid precipitation (see the 3.1.6section of the paper) makes it possible to estimate the ratio of these components it the total river flow.A series of articles was published from the beginning of 1970 s , describing runoff hydrograph separation by nourishment sources with the use of 18 O and D (see for example Dincer et al., 1970, Martinec et al., 1974, Fritz et al., 1976, Hermann et al., 1978, Cable, 2011).
A mixing-model approach was tested for the Djankuat River to conduct river hydrograph separation in the study (Rets et al., 2017).Two equation systems were drawn:1) in terms of water routing with salinity as indicator; 2) in terms of runoff genesis with δ 18 О as a tracer.In terms of water routing the Djankuat River hydrograph was separated into surface routed and subsurface routed waters.In terms runoff genesis the Djankuat River hydrograph was separated into liquid precipitation and meltwaters.Some 70% of the Djankuat River runoff in August-September 2014 was formed by ice and firn meltwater.Rain water is mostly subsurfacely routed; surface runoff of liquid precipitation is formed only during the most intensive rainfall (more than 20 mm on average).Ice and firn meltwater partly percolates to the glacier bottom and comes through a sub-surface layer.The fast (responsive to weather fluctuations) and regulated components of sub-glacier runoff can be distinguished.Sub-glacier runoff contributed 20-30% to the Djankuat river melt runoff in August 2014 and up to 100% of the Djankuat river melt runoff at the end of September 2014, when ablation stopped (Rets et al., 2017).

Sampling snow, ice, firn, liquid precipitation and groundwater in the basin for stable isotopes and electrical Conductivity
Regular sampling of snow, ice, firn, liquid precipitation and ground water in the Djankuat basin was carried out during 2014-2017.The first test sampling was performed in 2013.Liquid precipitation was sampled on every significant occasion of rainfall (more than 1 mm) that amounts to 25-30 samples a year of the Base camp weather station (Table 1).The snow, ice and firn sampling points were evenly distributed on the Djankuat basin area, the coordinates are given in the database.Samples were taken regularly during the ablation season, 40-150 samples a year.Snow samples were taken from the surface and on different depth of snowpack.The groundwater was sampled: 1) after the end of ablation season in the Djankuat River stream when the total runoff is assumed to be provided by groundwater in 2014; 2) out of the sub-glacial waters spring in 2015; 3) A groundwater-fed spring on the slope of the Djankuat River basin in 2017.
The values of δ 18 O and δD were measured in each sample.The proceeding of the samples was the same as for the stable isotopes samples taken at the gauging station (see section 3.1.5).Electrical conductivity was measured in the samples of snow, ice, firn and groundwater with conductometer Econics Expert-002.Total amount of samples taken and analyzed is: 113 samples of liquid precipitation, 218 samples of snow, 116 samples of ice, 22 samples of firn and 16 samples of groundwater (Table 4).
Relative concentrations of 18 О and D in precipitation are strongly correlated and defined to a great extent by air temperature, which is called the "seasonal isotope phenomenon" (Dansgaard, 1964).The highest concentrations of 18 О and D among the samples collected in the Djankuat River basin are characteristic of liquid precipitation (Table 4, Fig. 12).The δ 18 О value lies mostly between -0.5 and -7‰ with a mean value of -4.9‰,The δD value -between 0 and -40‰, the mean value is -26‰.The lowest concentrations of 18 О and D were registered in winter snow (δ 18 О=-28.3‰,δD=-216‰) and Djankuat glacier ice (δ 18 О=-22.0‰,δD=-159‰).The amount of stable isotopes in snow cover formed during spring snowfalls is higher and closer to the corresponding values in liquid precipitation (δ 18 О=-9…-5‰, δD=-28…-70‰).The mean concentration of 18 О and D is quite similar: the mean δ 18 О is -12.2‰ in snow samples, -14.3‰ in ice samples and -11.3‰ in firn; mean δD is -85.5‰,-99.3‰ and -77.1‰ respectively.The surface of a glacier ablation area is always formed by ice layers of different ages, while firn samples represent the climatic conditions of the last 5-10 years.A generally warmer mean isotopic composition of firn and snow compared to ice can indicate a total warming of the climate.Groundwater is a mixture of all the stated above sources, accordingly the points of groundwater samples lie in the middle of the δ 18 О vs. δD graph (Fig. 12).The mean concentration of precipitation in the replenishment of groundwater layers.
The ice, snow and firn samples are ultra-fresh (Table 4).Groundwater is enriched with dissolved salts up to 105 mg/L (114 μSm/cm).

Glaciological measurements
Glaciological observations were carried out by standard methods during 2007-2017 (Østrem andBrugman, 1992, Boyarsky, 1978).The data on snow depth, snow density and ablation is included in the presented database.
Snow depth on the Djankuat glacier is measured by probe poles at 250-300 points evenly distributed in all zones of the glacier (Fig. 13).The snow measurement survey usually starts in late May -early June and ends at the end of June.The total amount of 2932 measured values during 2007-2017 was included in the database (Table 5).The mean value of measured snow thickness is 3.6 m, the maximum is 11.5 m.
Snow density is measured in 2-4 snowpits placed in different elevation belts of the Djankuat glacier (Fig. 1).Density is measured in each 40-50 cm layer of a snowpack by a snow sampling cylinder.The measurements are repeated 2-5 times during the ablation season.Total amount of 66 measurements of integral density of snowpack and 434 measurements of density in layers of the snowpack were done during 2007 to 2017 (Table 5).Integral density of snowpack has a low variance, total range of variation is less than 0.2 g/cm 3 .The density in the layers of snowpack can greatly vary -from 0.23 to 0.92 g/cm 3 according to the 2007-2017 measurements.The overall mean measured value of snow density in the snowpack, 0.57 g/cm 3 , greatly exceeds the density of fresh snow, as the database includes measurements carried out in the midsummer and at the end of the ablation period (Table 5).
Ablation is measured by means of ablation stakes.35-45 stakes were placed on the Djankuat glacier surface every year.The time-step between measurements depends on the accessibility of each stake and ranges from 1-5 to 30 days.The values included in the dataset are counted from measured depth of melted snow/firn/ice in cm and corresponding values of density of melting material.Total amount of 5045 measurements of ablation was done during 2007 to 2017.The mean rate of snow/firn/ice ablation was 47 mm w.e./day with a minimum of 0 mm w.e./day and a maximum of 251 mm w.e./day.
The Djankuat glacier has experienced a general mass loss since the beginning of observation in 1968 (www.wgms.ch).Until 2005, negative mass balance years alternated with positive mass balance years (Fig. 14).After 2006 and during all period that is presented in the dataset, the annual mass balance of Djankuat glacier was negative.The annual mass balance values ranged from -2010 mm w.e. in 2007, that was the lowest value since the beginning of the observation in 1968, to -230 mm w.e. in 2009.The mean value of annual mass balance during 2007-2017 was -900 mm w.e.During 2007-2017 the area of the Djankuat glacier decreased from 2.68 to 2.42 km 2 that amounts to almost 10% of loss (Figure 13).The front of the glacier retreated by 60-300 m in different measurement profiles during 2010-2016 (Fig 13) (Fig 13), that amounts to 8.5-42.9m/year retreatment rate.The main reason of the Djankuat glacier retreating during the long-term period is a decrease in summer balance, while accumulation shows a statistically insignificant positive trend (Fig. 14).
Mass Balance Bulletin, which is issued by the World Glacier Monitoring Service (wgms) at 2-year intervals, and in Fluctuations of Glaciers along with other standardized data on changes in glaciers throughout the world at 5-yearly intervals (see for ex.Popovnin, 2012Popovnin, , 2013)).Some of the glacial measurements data presented in this article was used in global studies on evolution of the Earth's cryosphere (Zemp et al., 2009, Zemp et al., 2011, Zemp et al., 2015).In Popovnin and Pylayeva (2015) snow thickness measurements on Djankuat glacier are used to work out a methodology of estimation of avalanche feeding of a glacier from the total mass of snow accumulation.

Meteorological measurements
The main purpose of meteorological measurements in the Djankuat research basin is to provide the data needed for calculating the components of the heat balance (Toropov et al., 2017(Toropov et al., , 2018b)), which is a necessary input for physically-based hydrological models.Aa an example, the presented meteorological data was successfully used to model the melting regime of the Djankuat glacier in 2007 by the A-Melt model of ice and snow melt in alpine areas (Rets, Kireeva, 2010).
The program of meteorological observations in the Djankuat research basin during 2007-2017 included (Table 6): 1. Meteorological and radiation measurements above ice surface by means of Campbell AWS (Fig. 15a), including measurements of air temperature and relative humidity (Vaisala MT300 sensor), wind speed and direction (Campbell wind sensor) at 2 m AGL; radiation fluxes (KEEP & ZONNEN radiometers -two of them measure incoming and outgoing short-wave radiation, another two upward and downward long-wave radiation); measurements of ablation using a Sonic Ranger sensor (the sensor is located on a construction, that is drilled into the body of the glacier, and measures the distance from the sensor to the ice or snow surface).These automatic measurements have a record interval of 15 min.The weather station was placed in the central part of the Djankuat glacier (Djankuat Glacier AWS 1 on Fig. 1, Table 1) 2. The second Campbell AWS with the same parameters was placed in the central part of the glacier over the debris-covered surface in 2007, 2008 and 2009 (Djankuat Glacier AWS 2 on Fig. 1, Table 1).
3. Gradient mast DAVIS placed in the central part of the Djankuat glacier (Djankuat Glacier AWS 1 on Fig. 1, Table 1) includes 4 temperature and humidity sensors and 4 wind sensors located at 0.25, 0.5, 1 and 2 m above the glacier surface (Fig. 15c, Table 6).Measurements were recorded with a 15 min interval.
Observations were carried out in 2015 to obtain long-term meteorological data series in the surface layer, to obtain turbulent heat fluxes estimations with the Monin-Obukhov method (Zilitinkevich, S.S., 1972).
4. Measurements of turbulent pulsations of wind and acoustic temperature with a 3-axis sonic anemometer GILL WindMaster (Table 6) in the central part of the Djankuat glacier (Djankuat Glacier AWS 1 on Fig. 1, Table 1).The measurement frequency is 10 Hz.This measurement method yields fluxes of turbulent heat, moisture and momentum by the «eddy-covariance» method (Andreas et al., 2005).
5. Measurements of the basic meteorological parameters, such as air temperature, precipitation, atmospheric pressure, wind speed and direction, in the base camp area at an altitude of 2640 m.a.s.l.(Base Camp AWS on Fig. 1, Table 1) by a Davis meteorological station.These automatic measurements also have a record interval of 15 min.6.In 2017 a Davis AWS was also placed in the upper part of the glacier (Djankuat Glacier AWS 3 on Fig. 1, Table 1).The station worked with a record interval of 15 min (Table 6).
The Figure 16 shows an example of the course of the average daily values of the basic meteorological variables during the ablation season of 2007 measured by Campbell AWS 1.It is clearly seen that changes in air temperature associated with synoptic events are expressed quite well, their average amplitude is 3 °C (the same values were observed the other years).The average temperature during the ablation period is around 8 °C, while the minimum values almost reach 0 °C annually, and the maximums are 16 to 18 °C.The variability of the radiation balance is determined mainly by cloudiness, which has primarily a pronounced daily variation.The albedo effect is also clearly manifested -especially in June and September, when fresh snow frequently falls on the surface of the glacier.The maximum values of incoming shortwave radiation can reach 1100 W/m 2 .
The wind regime is stable to a great extent and varies little from year to year.The average wind speed over the tongue of the glacier is fairly stable at 4 m/s, while the maximum does not exceed 12 m/s.Above the glacier, stable katabatic winds blow, which is characterized by a pronounced diurnal course.The maximum speed values are associated with foehn winds, observed 3-4 times per season.Table 7 shows the average July values of the main meteorological characteristics.Figure 17 gives an example of the variability of the main meteorological characteristics in the Adyl-Su river valley, in the area of the base camp at an altitude of 2640 meters above sea level (Table 1, Fig. 1).The average daily temperature is about 12 °C, the average wind speed is about 5 m/s, despite the fact that the maximum gusts are stronger than over the glacier and reach 18 m/s.This is due to the density difference between the cold air flowing from the glacier and the local air mass forming over the heated alpine meadows and rocks.Heavy rains can be observed in the study area.For example, in 2017, the daily precipitation of about 20 mm was observed 6 times.From August 31 to September 1, about 48 mm of precipitation fell within 48 hours, which is a very high amount.This rainfall caused the above mentioned outburst of the Bashkara glacial lake in the neighboring valley and the formation of a catastrophic mudflow (Chernomorets et al., 2018).

Conclusion
With the detailed measurement program described here, the Djankuat basin is now a unique research site not only for the high elevation territories of the North Caucasus, but for the Russian Federation as whole (Konovalov et al., 2018, Stokes et al., 2006, Shagedanova et al., 2005, Hagg et al., 2010).The aim of the complex monitoring in the Djankuat basin is not only to fill a "blind-spot" in extremely underreported North Caucasus alpine territories but to provide data for detailed studies of hydrometeorological processes in mountain areas (Rets et al., 2017, Toropov et al., 2017).
The dataset presented here covers the period of 2007-2017 and can be useful to researchers developing and verifying hydrological, glaciological and meteorological models for mountainous territories, studying the recent climate and its impact on the cryosphere and hydrology, using isotopic and hydrochemical approaches to study the source areas of runoff.

Acknowledgements
This work was supported by the Russian Foundation for Basic . The second Davis (AWS3) station was placed on the upper part of the Djankuat glacier in 2017.A DAVIS Gradient mast was placed in the central part of the Djankuat glacier (AWS1) in 2015 to obtain long-term meteorological data series in the surface layer.Turbulent pulsations of wind and acoustic temperature were measured in 2013, 2014 and 2016 with a 3-axis GILL WindMaster sonic anemometer in the central part of the Djankuat glacier (AWS1).Hydrological measurements at the Djankuat gauging station (see Fig. 1, see temperature has a close to a uniform distribution on the duration curve and is mostly within the range of 1.2-4.5 ⁰C (Fig 8c).Water temperature of the Djankuat River has a great diurnal variation up to 4⁰C.Diurnal maximum of temperature is usually observed at day time before the beginning of an intensive rise of diurnal wave of meltwater flow.Mean daily value of water temperature generally rises trough the ablation season (Fig. 10).The maximum value (6.63 ⁰C) was registered on 18 th of September 2016 at midday.The minimum values (0.1 ⁰C) are observed during the night in the beginning of the ablation period (Fig 10).
In 2014 to 2016 all samples were processed in Saint Petersburg State University Resource center for Geo-Environmental Research and Modeling (GEOMODEL) on Picarro L-2120i.In 2014 50 control sampless were processed independently by two laboratories: a) Saint Petersburg State University Resource center for Geo-Environmental Research and Modeling (GEOMODEL) on Picarro L-2120i; b) the Stable Isotope Laboratory of the Geography Department of Lomonosov Moscow State University on a Finnigan Delta-V mass spectrometer.The difference in the definition for the same sample did not exceed 0.2‰.In 2017 the samples were measured at Climate and Environmental Research Laboratory (CERL) of Arctic and Antarctic Research Institute on laser analyzer Picarro L2140-i that uses Cavity Ring-Down Spectroscopy (CRDS) technique to define the δD and δ 18 O ratios in water samples.After each 5 samples we measured our work standard "SPB" (distilled Saint Petersburg tap water) calibrated against the IAEA standards, in order to obtain true values of the samples' isotopic composition.23 % of randomly chosen samples were re-measured in order to estimate the reproducibility of the results, accordingly to 0.06 per mil for δ 18 O and 0,4 per mil for δD, which is 2 orders of magnitude less than the natural variability of the isotopic composition of the studied samples.The values of δ 18 O and δD in the Djankuat River waters have a relatively even duration curve (Fig 9 e,f).The value of δ 18 O stays in range of -13.5…-11.5‰most of the time, δD in --95…-80‰.The mean values of δ 18 O and δD are correspondingly -12.5‰ and -86.2‰.Concentration of 18 О and D decreases with an increase in share of ice and firn melt in total river flow as shown in the beginning of June and July-August 2017 in Figure 10.Pronounced rises in δ 18 О and δD are driven by precipitation events (Fig 10)

Figure 1 :
Figure 1: The Djankuat river basin with the depicted location of the Base camp, main weather stations, snowpits and the Djankuat river gauging station.

Figure 2 :
Figure 2: A general view over the Djankuat research basin (photo by E.Rets).

Figure 3 :
Figure 3: Monthly distribution of air temperature and precipitation for the period of the dataset (2007-2017) compared to the long-5

Figure 4 :Figure 5 :Figure 6 :
Figure 4: The fluctuations of the mean air temperature: (Mean) annual, (MJJAS) from May to September, (ONDJFMA) -from October to April -according on the nearest to the Djankuat research basin all-year weather station Terskol (2146 m).The linear trends are shown for the period 1978-2017 which is identified as a "contemporary period" in term of the recent climate forced changes in river runoff regime on the European territory of Russia (Frolova et al., 2014, Rets et al., 2018).5

Figure 7 :Figure 8 :
Figure 7: An example of the Q=f(H) rating curve.Derived for the Djankuat Gauging Station for the 2016 ablation season

Figure 9 :
Figure 9: Fluctuations of mean monthly and mean seasonal runoff of Djankuat river during the whole period of observation (2007-2017).

Figure 11 :
Figure 11: Dependency between water electrical conductivity and Salinity for the Djankuat River basin.

Figure 12 :
Figure 12: δ 18 О vs. δD graph for the samples of snow, ice, firn, liquid precipitation and groundwater collected in the Djankuat River basin in 2013-2017, plotted with a global and local meteoric water line.

Figure 13 :
Figure 13: The spatial distribution of ablation stakes and snow thickness measurement points, change in the Djankuat glacier area during the study period and glacier front position.

Figure 14 :
Figure 14: Fluctuations of the Djankuat glacier mass balance components since the beginning of the observations in 1968 5

Figure 15 :Figure 16 :
Figure 15: Meteorological measurements on the Dzhankuat glacier at AWS-1: a) Campbell meteorological complex with a set of Kip & Zonen radiometers b) GILL three-component acoustic anemometer c) gradient mast equipped with temperature-humidity and wind sensors Davis (photo by M. Aleshina).5

Figure 17 :
Figure 17: An example of temporal variability of average daily temperature, wind speed and precipitation sum on base camp (point AWS-«base camp»): RAIN -precipitation, mm/day, T2m -air temperature, ⁰C, WS -wind speed, m/s.

Table 1 )
started from measuring runoff with onehour step during 2007-2010 ablation seasons.In 2013 the first test measurements of water conductivity, water salinity, water turbidity, and stable isotopes ( 18 O and 2 H), as well as first samplings of liquid precipitation, snow, ice, firn in the basin on water conductivity, water salinity, δ 18 O and δ 2 H were carried out.Since 2014 up to 2017 the stable isotopes sampling, conductivity measurements were done on a regular basis on the Gauging Station and on the watershed.A total amount of 844 samples on stable isotopes were collected on the Djankuat River Gauging station and 485 samples of snow, ice, firn, groundwater and liquid precipitation.Regular monitoring of the Djankuat River water turbidity (5-7 times a day) started during 2015-2017, of water conductivity -during 2014-2017, water temperature -during 2015-2017.