An 11-year ( 2007 – 2017 ) soil moisture and precipitation dataset from the Kenaston Network in the Brightwater Creek basin , Saskatchewan , Canada

Soil moisture and precipitation have been monitored in a hydrometeorological network situated within the Brightwater Creek basin, east of Kenaston, Saskatchewan, Canada, since 2007. The majority of the prairie landscape is annually cropped with some sections in pasture. This agricultural region is ideal for remotesensing validation and calibration and, in conjunction with the flux tower situated within the network, hydrological model validation. Remote-sensing validation collaborations have included the European Space Agency’s Soil Moisture and Ocean Salinity (SMOS) and NASA’s Soil Moisture Active Passive (SMAP). The network was developed at two spatial scales, one high-resolution set of sites installed over a 10 km× 10 km region and a second installed over 40 km× 40 km. The sites are all similar in design with three instrument depths for soil moisture and temperature, as well as precipitation measurement. The 2007–2017 dataset published in this paper has gone through a quality control review process, which involved both automated and manual processes. The dataset is limited to the summer months (1 May–30 September) due to the uncertainties and complexities of measurement in frozen soils and the freeze–thaw period each year. Data discussed in this publication are available at https://doi.org/10.20383/101.0116, and data beyond 2017 can be requested from the corresponding author.


1
Introduction 20 Soil moisture and precipitation are important elements of hydrological cycle. While it constitutes a small portion of the global water cycle, soil moisture has a significant influence on atmospheric and hydrologic processes.
Soil moisture is highly variable across a landscape, being influenced by both atmospheric conditions (e.g. precipitation, evaporation), landscape variability (e.g. topography, soil characteristics), and vegetation. This creates difficulty when attempting to asses soil moisture at the typical scales of atmospheric circulation models 25 (Crow et al 2012), however inclusion of soil moisture as a dynamic parameter within numerical modelling improves forecast skill for both hydrological and meteorological models (Koster et al. 2010;Koster et al. 2011;Drewitt et al. 2012;Wanders et al 2014). The difficulty of measurement has prompted researchers to develop remote sensing techniques to try and quantify soil moisture conditions at various scales. Any remote sensing technique requires calibration and validation, in this case achieved with in situ monitoring stations. 30 The Kenaston Network was designed to fulfil both the needs of land-atmospheric modelling and remote sensing validation programs. Very few existing monitoring networks have the ability to validate remote sensing products and hydrometeorological models over the Canadian prairies due to the unique combination of landscape and climatic conditions. Specifically for remote sensing of soil moisture, the individual stations were distributed at two spatial scales to accommodate validation of remote sensing products at various scales. The high resolution 35 of the network sites allows for both intergrid and intragrid validation.

Network Description
The Kenaston Network, also called the Brightwater Creek Monitoring Network is located on the Canadian Prairies in central Saskatchewan, approximately 80 km south of Saskatoon. Stations within the network consist of a series of soil moisture and precipitation sites, set at two spatial scales, and a year-round eddy-covariance 40 tower with a full complement of meteorological instrumentation. The monitoring sites are situated within the basin of Brightwater Creek, which drains northward into the South Saskatchewan River. Brightwater Creek has been monitored by a Water Survey of Canada flow gauge since 1965. The landscape is a typical agricultural region with annually cropped fields, mainly of cereals, oilseeds, and pulse crops, and pasture lands. The area is flat with slopes of less than 2%  which affects runoff in the region. Significant portions of the 45 area are considering non-contributing, where in general water does not drain to streams or rivers but instead ponds in small wetlands and sloughs (Shook et al., 2013). Predominantly silt loam, the area ranges from sandy loam to clay in texture.
Data from the network have been used for several projects over the years including the European Space Agency's (ESA) Soil Moisture and Ocean Salinity (SMOS) mission, the National Aeronautics and Space 50 Administration's (NASA) Soil Moisture Active Passive (SMAP) mission, the Drought Research Initiative (DRI), and the Changing Cold Regions Network (CCRN). A field campaign for the SMAP satellite was conducted in 2010 (CanEx-SM10), and previous publications that describe this network include Magagi et al. (2013) , Champagne et al. (2010;, Rowlandson et al. (2015), and Burns et al. (2016). Canada and the University of Guelph.

Soil moisture and precipitation site details
The soil moisture and precipitation sites are distributed at two spatial scales: 10 km 2 and 40 km 2 ( Figure 1). The larger scale network has been modified over time and began in a 45 x 55 km area. Each site consists of a datalogger, power system, tipping bucket rain gauge (TBRG), and 3-4 Hydra Probes. These sites are usually set 65 outside of the actively managed area of the field, in fence line strips, under powerlines, or at the very edge of the field. Figure 2 shows a typical setup. All sites have at least three probes, inserted horizontally at depths of 5, 20, and 50 cm below the surface that remain in place throughout the year. Additionally, site at the 10 km 2scale also have a vertically placed probe, generally indicated as 0-5 cm, which moves into the field after seeding and is removed shortly before harvest. Figure 3 illustrates the general setup of the stations that include a vertically 70 place probe, indicating the location of the four probes. Stations with only three probes have a similar setup, with all three probes inserted near the datalogger box.
Data is collected at 30 minute intervals, a single point measurement from each Hydra Probe and the sum over the interval for the TBRG. Provided from each probe for this dataset is real dielectric constant (real dielectric permittivity, ε r ), temperature, and soil moisture using the manufacturer's loam calibration equation. Additional 75 data has been collected at some sites within Kenaston, including soil conductivity, 2.5 cm soil temperature, crop types, heights, and photos, air temperature and relative humidity, point measurement snow depth, and snow surveys, which is not included in this dataset.
Sites are visited regularly throughout the field season to ensure TBRG cleanliness and check for site issues. Site with a vertically placed probe are visited more frequently than others due to the greater risk for disturbance and 80 placement issues.

Soil Instrumentation
The instrument used throughout the network to measure soil parameters is the Stevens Hydra Probe II (Stevens Water Monitoring Systems (Inc, 2009). These are radiometric coaxial impedance dielectric reflectometer sensors, with four tines extending from a 4.2 cm diameter head, along which a radio frequency is 85 applied and the reflected frequency measured (Stevens Water Monitoring Systems, Inc., 2018b). This reflected signal is related to the real dielectric constant (ε r ) of the soil which in turn is correlated to soil water content (e.g. Topp et al., 1980;Campbell, 1990;Seyfried et al., 2005). General ranges for ε r are roughly 80 in water, 1 in air, and 2-5 in dry soil. A more detailed description of the instrument and the measurement principles can be found in publications from Stevens Water Monitoring Systems, Inc. (2018a, 2018b). These sensors are widely used in 90 university and government research networks, including NOAA's Climate Reference Network (Bell et al., 2013), the USDA's Soil and Climate Analysis Network (Schaefer et al., 2007), and Agriculture and Agri-Food Canada's national monitoring networks (Adams et al., 2014).
Real dielectric constant (ε r ) is related to soil moisture through a calibration equation. The equations supplied from the manufacturer report a sensor accuracy of ±0.03 m 3 m -3 (Stevens Water Monitoring Systems, Inc., 2018a 95 or b), and a site specific calibration is recommended (e.g. Huang et al., 2004;Seyfried and Murdock, 2004;Rowlandson et al., 2013). The uncertainty in calibration method and ongoing work in this area presents a difficulty that has not been satisfactorily resolved, particularly for the measurements at deeper depths, as described in Burns et al. (2014). To ensure consistency for all of the data the manufacturer supplied loam calibration equation (Stevens Water Monitoring Systems, Inc., 2018b) is used to calculate soil moisture, with the 100 understanding that this decreases the overall accuracy of the network. In situ calibration equations have been established for the majority of the near surface probes (5cm) and these equations are available upon request.
Occasional measurement issues with the Hydra Probe were encountered, some of which may be specific to the Kenaston network. For example, during hot summer days when the surface soil becomes very dry, ε r from the near surface probes (vertically placed 0-5 cm and horizontally placed 5 cm) will drop below ~2.6968, which 105 produces a negative soil moisture value using the loam equation. These low ε r values are possibly due to soil cracking, poor sensor contact with the soil, or are simply valid responses from the probe. During these dry periods repositioning the probe, which is the typical response to these types of issues in near-surface probes, is not typically possible simply due to the difficulty in inserting a probe into dry, hard-packed, fine grained soils.
New cracks often form as the probe is taken out and re-inserted, resulting in the same issues. These probes are 110 closely monitored and after the next sufficiently significant rain event, soil moisture typically increases and the probe begins responding as expected. Additionally, a diurnal oscillation of measured ε r is observed, with greater amplitude during hot, dry conditions. This suggests a temperature effect on ε r but is not investigated further here (Seyfried and Grant, 2007).
The Kenaston region is similar to other parts of Saskatchewan in the occurrence of saline soils, the results of 115 which cause some issues with the deeper probes (horizontally placed probes at 20 and 50 cm) (Seyfried and Murdock, 2004). While a typical variation between successive timestamps outside periods of rainfall could be on the order of ±0.01 m 3 m -3 , those probes measuring in saline conditions can vary as much as at ±0.10-0.20 m 3 m -3 . This is corroborated by measurement of soil conductivity: increasing variability between consecutive timestamps coincides with an increase in conductivity, generally greater than 0.2 S m -1 . In some cases this only 120 occurs for a season, while other sites show a consistent record of high conductivity and therefore large measurement variation in soil moisture.

Precipitation Instrumentation
All sites within the network are equipped with a tipping bucket rain gauge (TBRG) to capture precipitation. One of two varieties are used: the Onset RG3 or the Hydrological Services TB3. All sites began with an Onset 125 TBRG but over the years they have been replaced within the 10 km 2 scale network to the configuration documented in Table 1. Currently all sites use a TBRG with a 0.2 mm scale but some earlier TBRG had a 0.1 mm scale. Common issues with the TBRG include blockage due to debris, mount damage from farm equipment, and the occurrence of single tips not related to network-wide rainfall events. Bird guards were installed on the TB3s where regular debris issues were common. Field calibrations of the TB3s have been regularly completed 130 since installation. A known issue with TBRG-style precipitation gauges is the possibility of single tips due to the retention of water in the bucket or siphon (the latter only in the case of the TB3). Single tips within the dataset that are not temporally correlated to a rainfall event may not be indicative of rainfall within the 30 minute measurement period. These records have not been removed from the dataset due to the uncertainty in consistently determining validity without removing significant credible data. 135

Quality Control Process and Data
While the network is currently run year round, at maximum only May 1 -September 30 is included for each data year. The main challenges are difficulties in measurement and calibration occur during the winter and shoulder seasons when the ground is transitioning between a frozen and thawed state (e.g. Williamson et al. control/quality assurance (QAQC) are performed to warm season data: an automated check and then manual review. The automated phase checks for logger errors and common sensor errors, with the secondary manual review process including a review of field notes and checks of all sensors for known instrument errors and gaps in the automated process. The automatic review begins with the raw measurements and can be completed in near real time, while the secondary manual review is completed on an as needed basis, or seasonally. 145

Automated Review Details
The automated review process checks for the limits documented in Table 2 and removes data outside of these thresholds. These checks mainly screen for obvious sensor errors and provide consistency for the next phase of QAQC. Also applied during this process are flags that are using during the manual process to check for common errors (Table 3). 150

Manual Review Details
After the automated process, a manual review of the resultant data is conducted to do a final review of the data from each instrument and each site. Hydra Probes are typically reviewed against the site's TBRG, to ensure that jumps in soil moisture correlate with precipitation events. The TBRG are reviewed collectively, as at least for the dense set of sites precipitation events will be collected by all instruments. This repetition of equipment 155 allows for a relatively high level of confidence in rainfall events and provides useful information to diagnose TBRG collection or measurement errors. Review of field notes and comparison of TBRG between nearby sites confirms TBRG cleanliness, (debris can delay or block rainfall passing into the buckets of the TBRG) and general agreement between sites. When disagreement between a single site and the majority is observed and confirmed by field visits, the data is removed. 160 Site visits can potentially cause erroneous data and the data from the day of each site visit is reviewed and edited for (1) extra TBRG tips due to cleaning; (2) erroneous data from the vertically placed 0-5 cm probe when it is moved into and out of the field; (3) other sensor issues that could result in incorrect data (physical damage, disturbance by field equipment or animals); (4) erroneous values from troubleshooting or maintenance checks.
These checks are done in conjunction with review of field notes. Data from each sensor is also visually plotted 165 and reviewed for general operation as sensor malfunction can often be caught in careful review of the sensor parameters. In this QAQC stage, the focus is on unexplained jumps or drops, gaps, and unusually high or low values that have not yet already been removed during the automated review. Any data diagnosed during this process as erroneous is removed from the final data set.

Summary
Data from 2007 -2017, May 1 -Sept 30, from the Kenaston Network in the Brightwater Creek basin in central Saskatchewan, Canada, has been quality controlled and compiled in a standard format. The network consists of 175 two scales of sites, each with 3 -4 Hydra Probes and a tipping bucket rain gauge. Included in this dataset from each Hydra Probe is soil moisture, temperature, and real-dielectric constant (ε r ). Some issues with the Hydra Probe have been identified and documented, and the overall network coverage is good. It is anticipated that this dataset will continue to provide useful information for remote sensing validation and calibration as well as hydrometeorological modelling efforts. hydrology models: evaluation and parameterization, Hydrol. Process., 27, 1875-1889, doi:10.1002/hyp.9867, 2013 Topp, G., Davis, J., and Annan, A.: Electromagnetic determination of soil water content: measurements in coaxial transmission lines, Water Resour. Res., 16, 574-582, 1980. Earth Syst. Sci. Data Discuss., https://doi.org/10.5194/essd-2018 Open       -60 < x < 60 Real dielectric constant (ε r , unit-less) 0 < x < 90 Soil moisture, loam calibration (VWC, (m 3 m -3 )) 0 < x < 1.0