A long-term hydrometeorological dataset ( 1993 – 2014 ) of a northern mountain basin : Wolf Creek Research Basin , Yukon Territory , Canada

A set of hydrometeorological data is presented in this paper, which can be used to characterize the hydrometeorology and climate of a subarctic mountain basin and has proven particularly useful for forcing hydrological models and assessing their performance in capturing hydrological processes in subarctic alpine environments. The forcing dataset includes daily precipitation, hourly air temperature, humidity, wind, solar and net radiation, soil temperature, and geographical information system data. The model performance assessment data include snow depth and snow water equivalent, streamflow, soil moisture, and water level in a groundwater well. This dataset was recorded at different elevation bands in Wolf Creek Research Basin, near Whitehorse, Yukon Territory, Canada, representing forest, shrub tundra, and alpine tundra biomes from 1993 through 2014. Measurements continue through 2018 and are planned for the future at this basin and will be updated to the data website. The database presented and described in this article is available for download at https://doi.org/10.20383/101.0113.

. This basin is operated by Environment Yukon, University of Saskatchewan and McMaster University with support from the Global Water Futures program and provides a long-term dataset for precipitation, air temperature, humidity, wind, radiation, soil moisture, soil temperature, streamflow, and snowpacks at multiple elevations. WCRB includes meteorological stations located from low to high elevation that include a special emphasis on measuring snowpack and snowfall, and discharge gauges to measure streamflow in the main outlet and several 5 tributary streams.
The diversity of the basin, combined with the available long term comprehensive hydrometeorological data, is responsible for the popularity of WCRB as a site to carry out cold regions research by scientists from across Canada and abroad. Road access in summer, local accommodation, a major airport and upper air station in Whitehorse, simple 10 winter logistics, data availability and ecological diversity make WCRB an ideal location for remote sensing validation and snow modelling activities. WCRB can be used for detailed modelling studies as it has been densely monitored, has sufficiently long observation records and extensive parameter measurements (Pomeroy et al. 1999;Pomeroy et al. 2003Pomeroy et al. , 2006McCartney et al. 2006;Carey et al. 2007;Dornes et al. 2008;Quinton and Carey 2008;MacDonald et al. 2009;Rasouli et al. 2014Rasouli et al. , 2018Rasouli, 2017).

20
The basin is composed of four monitored sub-basins of Upper Wolf Creek, Coal Lake, Granger Creek, and Lower Wolf Creek, one non-monitored sub-basin of Middle Wolf Creek, and also three distinct ecological biomes; alpine tundra (20%), shrub tundra and taiga (58%), and boreal forest (22%) (Figure 1). At the very highest elevations, bare rock, short tundra mosses, and grasses dominate land cover. Above the treeline, shrub tundra with dwarf birch and willow shrub heights from 30 cm to 3 m occupies the majority of the basin (Jorgenson and Heiner, 2004). The taiga 25 at middle to lower elevations is dominated by shrubs and sparse black spruce woodland. At the lowest elevations are lodgepole pine, white spruce, and trembling aspen forest stands (Francis et al. 1998). Lewkowicz and Ednie (2004) estimate that 43% of WCRB contains permafrost. The basin physiography is described in detail by Rasouli et al. (2014). Meteorological measurements are taken at three primary stations in each of the biomes (Table 1; Figure 1) and are described as:

30
(1) Alpine tundra station. A windswept, high alpine tundra plateau along the drainage divide at the northern edge just outside of the basin. Vegetation is sparse consisting of moss and lichens with scattered boulders up to 1 m tall. Sensors are mounted on a 3 m tower. Additional snow sensors are Geonor T200B and Nipher snowfall gauge.
(3) White spruce forest station. A dense canopy of mature white spruce forest (12-18 m) near the Wolf Creek basin outlet. Well suited for assessing the effects of forest cover on snow water equivalent (SWE). Sensors are mounted on a 21 m triangular tower at various heights below, in and above the forest canopy. The tower at this site was relocated in 1996 and caution should be used comparing data pre-and post-move. Additional snow sensors:

GIS Drainage Basin Descriptors
A digital elevation model, DEM with 30 m cell resolution was prepared by Dr. Lawrence Martz (Dept of Geography,

15
University of Saskatchewan). DEM, geographic information system files for streamflow gauges, meteorological stations, groundwater well, boundaries of the drainage basin and sub-basins, streamflow tributaries, and lakes are provided ( Figure 1). An image of land-cover is also presented.

4 Data description
Hydrometeorological data are available for WCRB for water years (WY) from 1993-1994 to 2013-2014 for the three main meteorological stations, one in each primary biome, and four streamflow gauges ( Figure 1; Table 1).
Measurements of snow depth and density along snow survey transects were collected at least monthly by Environment Yukon and university researchers at each of the three meteorological stations. These measurements provide model 25 diagnostic information in each biome (Pomeroy and Granger, 1999). High quality meteorological measurements of hourly air temperature, relative humidity, rainfall, wind speed, incoming and outgoing shortwave radiation, net radiation, soil heat flux, soil moisture and temperature, and daily precipitation observations for three stations (Table   2) and hydrological measurements of hourly streamflow data recorded by four gauges, snow water equivalent, snow depth, and water level at one groundwater well (Table 3) are presented for the period of 1993-2014 in this paper. Table   30 2 and 3 list the sensors measuring meteorology, streamflow, soil parameters, groundwater, and snow in the three sites in WCRB and provides the height or depth of the instruments, long-term climatological averages, and periods of observation record. All stations record data at 30 minute intervals, but are reported hourly (except for precipitation data that is reported daily) in this paper with continuous records for several parameters at the main stations. All stations Earth Syst. Sci. Data Discuss., https://doi.org /10.5194/essd-2018-118 Open Access ultrasonic snow depth sensors can contain substantial noise which is in part due to vegetation, falling snow particles, and movement of the sensor. As mentioned, the forest tower relocated in June 1996 so the first few years of data are similar but not directly comparable to the new site. Snow surveys have been carried out once a month and snow depth, 10 density, snow water equivalent (SWE) data have been collected over the period of 1994-2014 at three traditional snow survey sites at the alpine, and shrub tundra, and forest sites. Each has a 25 point transect with 5 density measurements.
Water level at streams was monitored continuously from ice thaw in early May to freeze-up in early October.
Discharge measurements involving manual stream velocity measurements throughout the year are taken 15 approximately once a month at Upper Wolf Creek, Coal Lake outlet, and Wolf Creek at Alaska Highway. Solinst leveloggers are currently used for measuring level at the four hydrometric stations (Figure 1). Other instruments that were previously used for measuring water level and streamflow are HOBO U20-001-04 pressure transducers, Ott Thalimedes shaft encoders and Leupold and Stevens A-71 Strip Charts. The Alaska Highway hydrometric station has the most field visits due to ease of access, has received the most attention due to its importance as the basin outlet, and 20 has bridge access for the measurement of exceptionally high flows. As a result, it has the highest quality discharge observations in the basin. However, at very high flows, the channel is not ideal, particularly on the left bank where a small stream enters the creek adjacent to the gauging well. Additionally, there is uncertainty with the staff gauge reading at high flows, due to hydraulic jumps. At the highest of flows, it may be best to rely on the instrument gauge readings. The Coal Lake station is located in a location with good channel geometry for rating development, yet has 25 been challenging to instrument early to capture high flows. Historically, at least one and possibly more outburst flood events have occurred which produce high discharge early in the season. At the very highest water levels experienced at the station, the site is susceptible to flooding. Granger Creek is a steep, narrow, rough, and turbulent creek that is not ideal for velocity-area measurements or rating curve development, and so there is likely considerably error associated with the site. A very high water discharge measurement has never been made at the site, and benchmarks 30 have only been in use since 2012 so water level data is not transferable from older data. The channel is well controlled so that flooding is not an issue. The flashy nature of high water at Granger Creek also makes it difficult to observe snowmelt induced peak flows, which generally occur towards late in the day. The Upper Wolf Creek station was partially buried by a debris flow caused by an intensive convective storm and flood in the summer of 2013 and has been inactive since then. The site has historically been a source of difficulty as it has already been relocated once due

Data processing and adjustments
Data loss has been a common problem during the early winter due to power failure at stations with the onset of darkness and cold weather. The Standpipe gauge at the shrub tundra site is a pressure sensing precipitation gauge that broadcasts real-time data every 3 hours. Nipher-gauge solid precipitation measurements were corrected using a wind undercatch correction equation (Goodison et al., 1998) with wind speeds measured from nearby gauge-height 15 anemometers. Precipitation data with poor quality were removed and gaps in data were filled by establishing regression equations for meteorological variables between each of the three meteorological stations and the Whitehorse WSO station, which is located 13 km from WCRB. The equations to fill the gaps in precipitation and to correct for the wind undercatch were listed in Table 4. Precipitation data from the Whitehorse Airport and Whitehorse Riverdale stations were taken to fill the gaps primarily in WY of 1996-1997. Changes in elevation rather than distance 20 between stations are assumed to be responsible for the meteorological variation in this study basin. First, total daily precipitation was separated to snowfall and rainfall phases using Harder et al., (2013). Then, the snowfall proportion of precipitation was corrected for wind undercatch in each site. When data was missing, precipitation at the shrub tundra or alpine station was estimated solely based on the Whitehorse WSO measurements. Whitehorse WSO snowfall measurements are collected using a Nipher and require appropriate wind correction using WSO wind speed 25 measurements. Whitehorse WSO precipitation was assumed equal to precipitation at the forest site because of their proximity and similar elevations. Using available monthly totals, a relationship was found between cumulative snowfall at the Standpipe ( ℎ ) and the Nipher gauges to calculate "Nipher equivalent" snowfall ( ℎ ) at the shrub tundra site as in Eq. (1). When Standpipe data were not available, a relationship between Nipher equivalent cumulative precipitation at the shrub tundra site and the Nipher precipitation at the Whitehorse WSO site ( ) was used as in
(2). The snowfall data at the shrub tundra site also require a Nipher wind undercatch correction. To estimate missing rainfall data in the shrub tundra site and the Standpipe gauge, a relationship between cumulative rainfall at Standpipe gauge ( ℎ ) and rainfall at Whitehorse WSO site was used as in Eq.
(3). To estimate missing snowfall at the alpine site a relationship was used for uncorrected cumulative snowfalls between the alpine and shrub tundra Nipher gauges based on available monthly data as in Eq. (4). This equations can also be applied to the calculated daily "Nipher shrub tundra equivalent" in Eq.
(1). To estimate missing rainfall data at the alpine site the relationship between cumulative rainfall at the alpine Geonor ( ) and the shrub tundra Standpipe (Eq. 5) was applied to the daily rainfall data calculated or measured for shrub tundra as in Eq.
(3). A relationship between cumulative rainfall data recorded at the shrub tundra Standpipe and Geonor was found as in Eq. (6).

5
For undercatch correction, gaps in wind speed data (U) also were filled by interpreting from other stations based on best fit linear equations. The catch ratio (CR) for the Nipher gauges was calculated from Eq. (7), in which wind speed in m s -1 is measured at the Nipher gauge height (Goodison et al., 1998). In order to calculate wind speed at gauge height, roughness lengths of 0.001 m at the alpine and 0.1 m at the shrub tundra were used. Regression equations were 10 used to fill the gaps in other meteorological variables in the three stations representing forest, shrub tundra, and alpine biomes in WCRB. The regression equations for air temperature, relative humidity, and wind speed were provided in Table 5. Air temperature data has undergone a thorough quality control process. This included comparing air temperature among the three sites to look for unrealistic differences. Where available, air temperatures were also compared between multiple measurements made at the same site. The gaps in hourly air temperature data that were 15 between 6-13% of the total data were also filled by regressions established between hourly temperatures in the three sites in WCRB (Table 5). When data is missing for a few hours, an interpolation between air temperatures before and after the gap is used.
The relative humidity (RH) data at the alpine site had periods of RH > 100% prior to correction. For the first few years 20 of the archive, the data often exceeded 100. From 1995-2005 the peak RH being recorded gradually increased each year. Starting in September 2005, the sensor was replaced with a new version, and data quality improved, rarely exceeding 100%. From 2010 to the present time the data was almost entirely below 100%. For a few years between 1998 and 2000, there was also a second RH sensor at the alpine site. This sensor confirmed that the main tower RH sensor was in fact providing values that were too high, but also showed that for low RH values it was also reporting 25 values that were too low. As such, an equation was developed to reduce the variation of the raw data ( ) using a second order polynomial as in Eq. (8) in which and are constant values (Table 4). The RH data at the shrub tundra and forest sites, which have similar patterns to the alpine site, were quality-controlled and unreasonable data were corrected.

30
The REBS instrument net radiation (Rn) measurements require wind speed corrections and are now understood to have considerable measurement uncertainty. REBS were operation prior to 2006. When wind speed (U) data was not available a generic equation as in Eq. (9) was applied and when wind speed data was available Eq. (10) is used (Table   4). In all cases, the propeller anemometers were the first choice used for wind speed, if not available, cup anemometers were used. The NR-Lite2 instruments also required a post correction in some cases. This correction is only required

7 Example of data use
Data over the period of 1993-2011 were used by Rasouli et al. (2014Rasouli et al. ( , 2018 and Rasouli (2017) to investigate the change in mountain snow and runoff regimes under simulated warmer and wetter conditions in an uncertainty 20 framework. Future projections from eleven of the regional climate models for the A2 scenario for the period 2041-2070 were used to perturb records of observations of temperature and precipitation in WCRB.

Final remarks
The data from the Wolf Creek Research Basin have contributed significantly to our understanding of snow and runoff processes in subarctic mountainous environments. The long-term dataset at multiple elevations and land-cover types can be used to examine the variability of hydrometeorological fluxes with elevation and land-cover. As an example, the data have been used to investigate the hydrological influence exerted by shrub tundra (Pomeroy et al., 2006). The 5 unique dataset will be valuable to research communities working on mountain and northern hydrology for various purposes such as hydrological model development, assessment of climate change impacts, and inter-site comparison of hydrological processes.

10
The