Data from: Spatial and temporal features of snow water equivalent across a headwater catchment in the Sierra Nevada

<p dir="ltr">Accompanying dataset for: “Spatial and Temporal Features of Snow Water Equivalent Across a Headwater Catchment in the Sierra Nevada”</p><h3>Abstract:</h3><p dir="ltr">Here, we present the accompanying snow water equivalent (SWE) dataset in support of the scientific publication entitled “Spatial and Temporal Features of Snow Water Equivalent Across a Headwater Catchment in the Sierra Nevada” by the same authors listed above, currently under review in the journal “Hydrology and Earth System Sciences”. The dataset includes distributed SWE for 48 dates between water years (WYs) 2013 and 2017 for the Tuolumne River Basin in the state of California, USA. Alongside the dataset, we present a full description of the data and methods, along with a README file listing the contents of the archive.</p><h3>1 Data and Methods</h3><p dir="ltr">The Tuolumne River Basin is a headwater catchment in the California Sierra Nevada (Figure 1) with elevations ranging between 1150 m and 3999 m, and with a contributing area of 1180 km² to the Hetch Hetchy Reservoir (Hedrick et al., 2018b). Tree line in the watershed is located at around 2900 m, with land cover dominated by conifer forest below this elevation, and exposed granite bedrock above. The watershed is dominated by north-west and south-east aspects (Figure 2a and b) consistent with the predominant orientation of the river network (generated using TauDEM from a 10-m digital elevation model (DEM); http://hydrology.usu.edu/taudem/taudem5/index.html). Elevations above 2000 m (90% of the area, Figure 2c and d) are snow-dominated (~70 % of annual precipitation (Hedrick et al., 2018b)), while the lower elevations are rain dominated and/or in the rain/snow transition zone, although such elevations are variable between storms (Lundquist et al., 2016). The contributing area is dominated by mid-elevations between 2500 m and 3250 m (64 % of the basin) (see journal publication for location and details), with 26 % below 2500 m and 10 % above 3250 m.</p><p dir="ltr">This snow water equivalent (SWE) dataset is the product of NASA Jet Propulsion Laboratory (JPL) Airborne Snow Observatory (ASO) lidar-derived snow depths (3-m resolution snow depths resampled to 50-m) and iSnobal modeled snow densities at matching 50-m resolution for water years (WY, defined as October 1 – September 30) 2013-2017. The methodology to produce the 50-m SWE dataset is described in Hedrick et al. (2018b, 2020) and were generated through a reproducible framework (Hedrick et al., 2018a). The process of deriving hourly 50-m forcing grids from station measurements of air temperature, relative humidity, wind speed, precipitation, and incoming solar irradiance is described in Havens et al. (2017). For the extreme snow year 2017 during which some weather stations were buried, the model was forced using weather and precipitation data from the High Resolution Rapid Refresh atmospheric model (HRRR, Benjamin et al., 2016) following the methodology by Havens et al. (2019). Vegetation data from the National Land Cover Database (https://www.usgs.gov/centers/eros/science/national-land-cover-database) was used to estimate canopy effects on net radiation and turbulent fluxes following the methodologies described in Link and Marks (1999b, a) and Marks et al. (2008) implemented in iSnobal.</p><p dir="ltr">Lidar-derived snow depths from ASO (Painter et al., 2016) were used to periodically update the model snow depth state variable using the direct insertion approach (Hedrick et al., 2018b), which leaves the remaining model state variables of density, snow temperature, and liquid water content unchanged. This approach allows for the correction of the snow distribution by adding spatial variability to the snow fields in the model, with additional improvements in the model physics representation. Furthermore, although the effects of mass transport due to avalanche or wind redistribution are not explicitly addressed in the model, the direct insertion of the lidar snow depths allows for the correction of where these deposits are located with each update. The uncertainty associated with the 50-m lidar snow depth ASO product has been estimated to be of the order of 0.08 m (Painter et al., 2016).</p><p dir="ltr">In iSnobal, density of newly fallen snow was estimated using storm dew point temperatures, after which bulk snowpack density was tracked using snowpack temperature gradients, overburden pressure, and the presence of liquid water within the ice matrix. The snow density model implemented in iSnobal follows the approach by Anderson (1968, 1976), which was based on the field and cold room measurements by Yosida (1963, 1958), Mellor (1964) and Kojima (1967). These same equations from Anderson (1976, 1968) were later updated by Oleson et al. (2013) within the Community Land Model (CLM) and were further adjusted to estimate bulk snow density before being incorporated into iSnobal. Comparisons at the point scale of simulated to in-situ bulk density at 11 validation sites in Idaho and California during the density model development and implementation showed Nash-Sutcliffe coefficients of 0.83 +/- 0.08, mean bias error of +13 +/- 14 kg m^-3 and root-mean square difference of 40 +/- 13 kg m^-3.</p><p dir="ltr">Evaluation of model densities at the gridded spatial scales (e.g., 50 m) are more challenging and have unquantifiable added uncertainties due to the discrepancy of scales between available measurements (e.g., snow courses, snow pillows and snow pits) and the spatial grid cell footprint and thus have not been performed. Note here that measurements from snow courses are often reported as the average of several snow core measurements along transects that can be more than a kilometer long (e.g., from the California Department of Water Resources (CDWR) through <a href="https://cdec.water.ca.gov/" target="_blank">https://cdec.water.ca.gov/</a>). Snow densities are often derived from these reported mean SWE and mean snow depth values for these transect/courses, resulting in uncertainties that cannot be directly quantified. Additionally, uncertainties are also associated with direct local measurements of snow density. For example, bulk densities from snow pit profiles can carry 10+% uncertainties and are seldom available in practical settings (Proksch et al., 2016). Similarly, snow depth sensors at snow pillow stations are not always located directly over the snow pillow, leading to uncertainties that are difficult to address, with errors in derived snow densities that become more pronounced when the snowpack is shallower (e.g., early or late snow season). This location offset between sensors has been confirmed at several snow pillow stations in California, including a station in the Tuolumne Basin.</p><p dir="ltr">The resulting dataset consists of distributed SWE covering the entire watershed at 50-m resolutions. There are a total of 48 survey dates during the period WYs 2013-2017, with flight intervals (days between flights) between 4 to 35 days, and with 6 to 13 flights per year (Table 1). The survey dates generally capture snow conditions around or just prior to peak snowpack storage and throughout the subsequent snowmelt seasons. In WY 2017, flights started in January and continued well into the accumulation season (please refer to the journal publication that this dataset supports for detailed discussions). The study period provides the unique opportunity to analyze some of the most extreme water years on record, with the extreme drought of WY 2015, followed by the near average WY 2016, and culminating with WY 2017, one of the snowiest water years on record (Ullrich et al., 2018; Wang et al., 2017).</p><p dir="ltr">References listed in references.txt or Related Materials. </p>

[
  "005:18"
]

[
  "005:040"
]

[
  "geospatial"
]