1. Introduction
Drought conditions and record low snowpacks in the western United States during water years (WY; 1 October–30 September) 2014 and 2015 provided incentive to study the emerging phenomenon known as snow drought (Cooper et al. 2016; Mote et al. 2016). The simplest definition of snow drought is near- or above-average accumulated precipitation (P) coinciding with below-average snow water equivalent (SWE) at a point in time, typically 1 April when the climatological maximum of SWE occurs (Pederson et al. 2011). Harpold et al. (2017) expanded this definition into two types of snow droughts: warm snow drought, where October through March accumulated P is greater than 100% of normal and SWE is less than 100% of normal on 1 April, and dry snow drought, where October through March accumulated P is less than 100% of normal and SWE is less than 100% of normal on 1 April. Beyond simple evaluations (e.g., Sproles et al. 2017), little attention has been given to the temporal evolution of snow droughts in terms of persistent dry spells or individual storm events. Understanding the hydrometeorological processes that create snow droughts will aid in evaluating their impacts on consumptive uses that depend on snowmelt-derived runoff or ecological processes that depend on the presence of a snowpack. At present, these specific impacts are not well characterized beyond broad knowledge that shifts from snow to rain reduce warm-season streamflow (Berghuijs et al. 2014) and that warming winters cause less efficient snowmelt (Barnhart et al. 2016).
We postulate that defining snow drought based upon single points in time (i.e., 1 April) may result in misleading evaluations of snow droughts by ignoring the mechanisms that lead to their onset. Our purpose is to show how eight snow drought years identified on 1 April in the northern Sierra Nevada (Figure 1) can have varying hydrometeorological origins through the use of observational data at monthly, daily, and hourly time scales. The results lead us to recommend that snow droughts identified by the simple definitions given by Harpold et al. (2017) are explored temporally to understand the root cause of the identified P to SWE divergence. We also recommend that continuous observations of snow drought be implemented in order to identify early and midwinter snow droughts that were later terminated by subsequent snowfall prior to 1 April. The goals of these recommendations are to improve subsequent studies focused on examining how snow droughts impact hydrologic, ecologic, and socioeconomic systems as well as identifying future snow drought likelihood in regions dependent on snow-derived water resources.
Map of the northern Sierra Nevada study area. Snow course dots with bold outlines were used in Figure 3.
Citation: Earth Interactions 22, 2; 10.1175/EI-D-17-0027.1
2. Data and methods
Monthly snow course data (https://wcc.sc.egov.usda.gov/nwcc/rgrpt?report=snowcourse) from 18 snow courses (Figure 1) and P estimates from the Parameter-Elevation Regressions on Independent Slopes Model (PRISM; Daly et al. 2008) were used to identify warm snow droughts based upon 1 April accumulated P > 100% and SWE < 100% of normal. Dry snow droughts were identified based upon 1 April accumulated P < 100% and SWE < 100% of normal. Each snow course had at least 80% of the 1981–2010 climatology period in the record. Monthly values of temperature (T) and P from the PRISM grid point nearest to three snow courses (Figure 1) spanning low (1997 m), middle (2103 m), and high elevations (2591 m) are included to examine the evolution of these snow droughts as a function of elevation. We define early-, middle-, and late-onset snow droughts based upon the time their onset was identified, with October–November characterizing early, December–February as middle, and late onset as March–April.
To explain the observed accumulated P/SWE ratios and explore impacts on streamflow, we included the following additional observations in our analysis: 1) daily, quality-controlled P, T, and SWE from three Snowpack Telemetery (SNOTEL) stations [Tahoe City Cross (2072 m), Squaw Valley Gold Coast (2442 m), and Fallen Leaf Lake (1903 m); Figure 1] acquired from the National Resources Conservation Service (https://www.wcc.nrcs.usda.gov/snow/snotel-data.html); 2) brightband-derived snow levels (White et al. 2010) from the NOAA Hydrometeorological Test bed/California Department of Water Resources (DWR)-supported snow level radar located at Colfax, California, which was acquired from the Earth Systems Research Laboratory (http://www.esrl.noaa.gov); 3) daily streamflow from the North Fork of the American River acquired from the U.S. Geological Survey (https://waterdata.usgs.gov/nwis/uv?site_no=11427000); 4) GPS-measured precipitable water from the Bodega Bay Atmospheric River Observatory acquired from SuomiNet (http://www.suominet.ucar.edu/; Ware et al. 2000); 5) P values from a DWR weather station at Blue Canyon (1610 m; http://cdec.water.ca.gov/); and 6) daily maximum T and accumulated P from five National Weather Service Cooperative Observer (COOP) stations located in the northern Sierra Nevada above 1300-m elevation with 90% complete records from October 1950 to February 2017 acquired from the Applied Climate Information System (ACIS) database (http://scacis.rcc-acis.org/).
We calculated the fraction of precipitation as snow (hereafter snow fraction) using the Dai (2008) equation, which estimates conditional probabilities of snow as a function of T with parameter values for the Sierra Nevada ecoregion (Rajagopal and Harpold 2016). We calculated the snow fraction for October–March (February in 2017) and took the five-station average over this period.
3. Results and discussion
3.1. A historical monthly perspective using snow courses
Seven snow drought years are identified from snow course data using the 1 April P to SWE criteria (Figure 2a). One is a warm snow drought (WY1951), two are influenced by singular wet events (WY1963) or wet periods (WY1997), two are late-onset snow droughts (WY1970 and WY2016), and two are dry snow droughts (WY1977 and WY2015). Figure 3 shows monthly evolutions of these snow droughts via PRISM estimates of T and P and first-of-the-month SWE measurements for three snow courses (bold outlined dots in Figure 1). WY1951 represents a classic warm snow drought. This year was characterized by late fall (October–November) having well above normal P and anomalously warm conditions followed by a wetter than normal winter (December–January) while late winter (February–March) was dry and cool. WY1963 demonstrates how a singular wet event can obscure an otherwise dry snow drought year, instead leading to its categorization as a warm snow drought. WY1963 began with an extremely wet October, resulting from the warm and wet Columbus Day Storm of 1962 and was followed by a dry and anomalously warm winter. The October event created early snow drought conditions with divergent P (above normal) and SWE (below normal). Wetter than normal conditions during February–March provided marginal SWE gains; however, the 1 April P to SWE ratio satisfied the snow drought criteria. WY1997 provides another example of how an extremely wet period (the 1997 flood; Lott et al. 1997; Kaplan et al. 2009) can facilitate producing a snow drought year. Average T during WY1997 did not appear to play a major role in producing snow drought conditions, although a dry and warm March reduced SWE at the low-elevation snow course (Figure 3g). This created a late-onset snow drought because of the above-average accumulated P. The high humidity and high snow levels during the 1997 flood event melted substantial snow (Underwood et al. 2009; Kaplan et al. 2009) and resulted in strong elevation dependence of warm snow drought conditions (Figure 2c). WY1970 and WY2016 (see next section) also appear as late-onset snow droughts. The limitation of using a monthly time step becomes evident when trying to assess the precise processes causing WY1970 to transition into snow drought. WY1977 is a classic dry snow drought where P and SWE were both much below normal. WY2015, noted for its exceptional lack of snow (Belmecheri et al. 2015) and persistent above-normal T (Figure 3), is classified as a dry snow drought caused by low P (Figure 3; Harpold et al. 2017).
(a) Scatterplot of PRISM October–March accumulated P and snow course SWE for 1 Apr during the period spanning WY1951–2017 as a percentage of 1981–2010 medians. Each dot represents a different snow course. All years are shown in gray with identified snow droughts colored. WY2017 (black dots) is also shown since, although it failed to be classified as a snow drought on 1 Apr, snow drought conditions were present during November–December. WY2017 is presented as a case study in section 3.3. (b) Percentage of 1 Apr 1981–2010 median snow course SWE vs elevation for WY1970. (c) As in (b), but for WY1997.
Citation: Earth Interactions 22, 2; 10.1175/EI-D-17-0027.1
Monthly evolution of (a) T, (b) P, and (c) SWE for seven snow droughts at Carson Pass (elevation 2591 m). (d)–(f) As in (a)–(c), but for Donner Pass (elevation 2103 m). (g)–(i) As in (a)–(c), but for Camp Richardson (elevation 1997 m). The T values are anomalies (°C), and P and SWE values are the percent of 1981–2010 climatology.
Citation: Earth Interactions 22, 2; 10.1175/EI-D-17-0027.1
3.2. Daily data reveals onset of the 2016 snow drought
Daily observations of P and SWE from SNOTEL stations provided information relevant to the late-season snow drought onset during WY2016. As previously shown in other water years, the influence of elevation on snow drought appears in snow course data (Figure 2c). In 2016, the elevation dependence of the percent of normal 1 April SWE is shown by above-normal conditions at the highest-elevation station but below normal at the lowest-elevation station (cf. Figures 3c and 3i). Daily data at the Tahoe City Cross station illustrate this further (Figure 4) as both accumulated P and SWE were near or above median values through February 2016. However, two dry periods with anomalously warm conditions during February (Figures 3a,d,g and 4a; right y axis) coincided with SWE depletion to below median values. This period marked the onset of snow drought at Tahoe City, with accumulated P above and SWE below normal. Although P accumulations continued in March and April, only marginal SWE recovery occurred because of the above-normal T during March (Figure 3). Using the normalized difference snow index (NDSI) values from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor on board the National Aeronautics and Space Administration Terra and Aqua satellites and acquired using the Google Earth Engine–driven Climate Engine (Huntington et al. 2017), the shift from an above-normal early season snowpack to widespread low-elevation snow drought (cf. Figures 4b and 4c) following the warm, dry period is shown spatially. Remotely sensed products such as NDSI may provide valuable data for real-time snow drought monitoring in conjunction with ground-based station observations.
(a) Time series of 1981–2010 median and WY2016 observed SWE, P, and T anomalies for the Tahoe City Cross SNOTEL spanning 1 Oct 2015 to 30 Apr 2016. MODIS-derived normalized snow difference index values (differenced from 2002 to 2017 averages) for (b) the 30-day periods during no snow drought and (c) after the onset of low-elevation snow drought.
Citation: Earth Interactions 22, 2; 10.1175/EI-D-17-0027.1
3.3. A case study of 2017
Daily and hourly observations add further insight to the physical processes driving onset and termination phases of snow droughts. During the early portion of WY2017, concern mounted as to whether WY2017 would become a warm snow drought year. Lack of snow during the Christmas–New Year’s period can have marked economic impacts on winter tourism in the northern Sierra Nevada. To examine the sequence of events leading to warm snow drought onset and termination between 1 October and 28 February 2017, we use daily P and SWE data from two SNOTEL stations: a higher-elevation station at Squaw Valley Gold Coast (2442 m; hereafter Squaw; Figure 5a) and a lower-elevation station at Fallen Leaf (1902 m; Figure 5b). Hourly observations (Figures 5c–e) are included to facilitate explanation of the processes driving snow drought variability.
(a) Median 1981–2010 P (salmon) and SWE (light blue) and WY2017 observed P (red) and SWE (dark blue) at Squaw Valley Gold Coast SNOTEL for 1 Oct–28 Feb 2017. (b) As in (a), but for Fallen Leaf SNOTEL. (c) Brightband-derived snow levels derived from the Colfax, California, snow level radar. (d) Streamflow at gauge on the North Fork of the American River. (e) GPS-derived total precipitable water (left) at Bodega Bay and precipitation (right) at Blue Canyon. Blue bars denote atmospheric river events.
Citation: Earth Interactions 22, 2; 10.1175/EI-D-17-0027.1
Early WY2017 featured multiple landfalling atmospheric rivers (Figure 5e; defined following Ralph et al. 2004) producing copious P with very high snow levels (>2500 m) and little snow accumulation. A dry November marked the onset of snow drought conditions (O1; brown boxes) at Squaw. A higher snow level storm in early December led to recovery at Squaw but produced warm snow drought onset (O1F) at Fallen Leaf. A moderate snow level atmospheric river event contributed substantial P with SWE increase at both SNOTEL stations leading to snow drought termination (T1F) on 3–5 January at Fallen Leaf. A higher snow level atmospheric river event during 8–10 January nearly brought Fallen Leaf back into snow drought because of the appreciable rain on snow. During the remainder of February, SWE at both elevations rose steadily, with one exception being another high snow level and wet atmospheric river on 8–9 February. No further periods of snow drought were observed.
WY2017 demonstrated the utility of employing daily and subdaily observations in terms of monitoring potential snow drought onsets and terminations and their root causes including prolonged dry spells and high snow level precipitation events. Snow level radar data showed how differing phases of P associated with heavy P events caused melting and accumulation of snow to vary with time and by elevation on a storm-by-storm basis. Atmospheric rivers, which provide large P totals and frequent rain-on-snow events caused by warmer temperatures and higher snow levels (Guan et al. 2016; Hatchett et al. 2017), appeared to contribute to snow drought onset and termination during WY2017. The correspondence between atmospheric river conditions, heavy P, and high snow levels with varying low-elevation SWE behavior and streamflow peaks (Figure 5d) emphasizes how high-frequency observations facilitate explanation of the hydrometeorological processes that contribute to snow drought variability. Examination of multiple stations at high-frequency intervals can also aid in identifying potentially impactful snow drought conditions, such as low-elevation snow drought (O1F in Figure 5).
3.4. Snow fraction and midwinter streamflow
Partitioning total P into snow fractions and examining midwinter peaks in streamflow gives additional process-based insight into the origins and implications of snow droughts. In six of seven snow droughts, the snow fractions (Figure 6a) are below the 5-yr moving average (thick black line on Figure 6a). Four of the lowest snow fractions observed over the period studied occurred during snow drought years. Four of the seven snow drought years coincided with peak midwinter runoff events in the North Fork of the American River (Figure 6b). This result is consistent with low snow fractions as rain-on-snow contributes to midwinter runoff and snowpack losses or minimal gains (Figure 5). Under future climate regimes (Klos et al. 2014) where stronger storms are likely (Lavers et al. 2015), the concern of shifting the P phase toward more rain (lower snow fractions) in T-sensitive mountain ranges such as the Sierra Nevada (Safeeq et al. 2016; Hatchett et al. 2017) implies that snow drought may become increasingly frequent, especially in lower-elevation mountains such as the northern Sierra Nevada. The WY2015–17 period may be an excellent test case for future climate regimes and how managers can adapt to them. During these 3 years, P covered the full range of hydroclimatic extremes (anomalously low to anomalously high), and 3 consecutive years of different flavors of snow droughts were observed: dry in WY2015, late onset in WY2016, and early season in WY2017.
(a) Fraction of October–March precipitation estimated as snow and averaged over five COOP stations. Black line shows the 5-yr moving average. Dots are colored by total October–March precipitation and sized proportional to the median number of precipitation days. (b) Daily observations of October–March streamflow at the North Fork of the American River. The dashed blue line shows the top 0.02% of streamflow from the period 1950–2017. Red boxes on both plots show identified snow drought years (cf. Figure 2), and the black box shows WY2017.
Citation: Earth Interactions 22, 2; 10.1175/EI-D-17-0027.1
4. Concluding remarks
While not a comprehensive evaluation of all historic snow droughts in the northern Sierra Nevada, our preliminary findings indicate that snow droughts have varied mechanistic origins. The temporal divergence of P to SWE can be used as a metric to estimate the onset of snow drought conditions (Sproles et al. 2017), particularly when examined in conjunction with anomalous T (Cooper et al. 2016; Harpold et al. 2017) and the characteristics of individual storms. These characteristics include snow levels during significant P events, persistence of warm and dry conditions, or singular storm events. Years that do not satisfy the snow drought criteria applied on 1 April (WY2017) should still be evaluated to identify early winter snow drought conditions and potential impacts to ecological processes (e.g., Campbell et al. 2005) and socioeconomic impacts (e.g., Harpold et al. 2017). Warm snow droughts frequently corresponded with water years that included lower fractions of total precipitation falling as snow and often included midwinter flood events, implying that rain-on-snow events often create warm snow drought conditions. We hope that the considerations presented herein will yield robust snow drought climatologies and assist in developing real-time snow drought monitoring that assimilates high-resolution observational data and in characterizing time-dependent snow drought impacts. The importance of submonthly data is highlighted by aiding the identification of specific processes producing (or terminating) snow droughts, such as winter heat waves or extreme precipitation events with high snow levels. It also facilitates understanding of potential impacts such as midwinter peak streamflow following rain-on-snow events. Future research should focus on identifying relevant time scales (durations and timing of onset) of snow drought impacts for hydrologic and ecological processes. For example, hydrologic modeling could identify threshold responses in watersheds to changes in precipitation phase and snowpack accumulation patterns relevant for water resource management.
Acknowledgments
The project described in this publication was supported by Grant G14AP00076 from the U.S. Geological Survey and the National Integrated Drought Information System of the National Oceanic and Atmospheric Administration under Grant AB133E-16-cQ-0022. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the USGS. This manuscript is submitted for publication with the understanding that the U.S. government is authorized to reproduce and distribute reprints for governmental purposes. We thank N. Oakley, S. Hatchett, and two anonymous reviewers for constructive comments that improved this manuscript.
References
Barnhart, T. B., N. P. Molotch, B. Livneh, A. A. Harpold, J. F. Knowles, and D. Schneider, 2016: Snowmelt rate dictates streamflow. Geophys. Res. Lett., 43, 8006–8016, https://doi.org/10.1002/2016GL069690.
Belmecheri, S., F. Babst, E. R. Wahl, D. W. Stahle, and V. Trouet, 2015: Multi-century evaluation of Sierra Nevada snowpack. Nat. Climate Change, 6, 2–3, https://doi.org/10.1038/nclimate2809.
Berghuijs, W. R., R. A. Woods, and M. Hrachowitz, 2014: A precipitation shift from snow towards rain leads to a decrease in streamflow. Nat. Climate Change, 6, 583–586, https://doi.org/10.1038/nclimate2246.
Campbell, J. L., M. J. Mitchell, P. M. Groffman, L. M. Christenson, and J. P. Hardy, 2005: Winter in northeastern North America: A critical period for ecological processes. Front. Ecol. Environ., 3, 314–322, https://doi.org/10.1890/1540-9295(2005)003[0314:WINNAA]2.0.CO;2.
Cooper, M. G., A.W. Nolin, and M. Safeeq, 2016: Testing the recent snow drought as an analog for climate warming sensitivity of Cascades snowpacks. Environ. Res. Lett., 11, 084009, https://doi.org/10.1088/1748-9326/11/8/084009.
Dai, A., 2008: Temperature and pressure dependence of the rain-snow phase transition over land and ocean. Geophys. Res. Lett., 35, L12802, https://doi.org/10.1029/2008GL033295.
Daly, C., M. Halbleib, J. I. Smith, W. P. Gibson, M. K. Doggett, G. H. Taylor, J. Curtis, and P. A. Pasteris, 2008: Physiographically-sensitive mapping of temperature and precipitation across the conterminous United States. Int. J. Climatol., 28, 2031–2064, https://doi.org/10.1002/joc.1688.
Guan, B., D. E. Waliser, F. M. Ralph, E. J. Fetzer, and P. J. Neiman, 2016: Hydrometeorological characteristics of rain-on-snow events associated with atmospheric rivers. Geophys. Res. Lett., 43, 2964–2973, https://doi.org/10.1002/2016GL067978.
Harpold, A. A., M. Dettinger, and S. Rajagopal, 2017: Defining snow drought and why it matters. Eos, Trans. Amer. Geophys. Union, 98, https://doi.org/10.1029/2017EO068775.
Hatchett, B. J., B. Daudert, C. B. Garner, N. S. Oakley, A. E. Putnam, and A. B. White, 2017: Winter snow level rise in the northern Sierra Nevada from 2008 to 2017. Water, 9, 899, https://doi.org/10.3390/w9110899.
Huntington, J., K. Hegewisch, B. Daudert, C. Morton, J. Abatzoglou, D. McEvoy, and T. Erickson, 2017: Climate engine: Cloud computing and visualization of climate and remote sensing data for advanced natural resource monitoring and process understanding. Bull. Amer. Meteor. Soc., 98, 2397–2410, https://doi.org/10.1175/BAMS-D-15-00324.1.
Kaplan, M. L., C. S. Adaniya, P. J. Marzette, K. C. King, S. J. Underwood, and J. M. Lewis, 2009: The role of upstream midtropospheric circulations in the Sierra Nevada enabling leeside (spillover) precipitation. Part II: A secondary atmospheric river accompanying a midlevel jet. J. Hydrometeor., 10, 1327–1354, https://doi.org/10.1175/2009JHM1106.1.
Klos, P. Z., T. E. Link, and J. T. Abatzoglou, 2014: Extent of the rain-snow transition zone in the western U.S. under historic and projected climate. Geophys. Res. Lett., 41, 4560–4568, https://doi.org/10.1002/2014GL060500.
Lavers, D. A., F. M. Ralph, D. E. Waliser, A. Gershunov, and M. D. Dettinger, 2015: Climate change intensification of horizontal water vapor transport in CMIP5. Geophys. Res. Lett., 42, 5617–5625, https://doi.org/10.1002/2015GL064672.
Lott, N., D. Ross, and M. Sittel, 1997: The winter of ’96–’97 West Coast flooding. National Climatic Data Center Tech. Rep. 97–01, 22 pp.
Mote, P. W., and Coauthors, 2016: Perspectives on the causes of exceptionally low 2015 snowpack in the western United States. Geophys. Res. Lett., 43, 10 980–10 988, https://doi.org/10.1002/2016GL069965.
Pederson, G. T., and Coauthors, 2011: The unusual nature of recent snowpack declines in the North American cordillera. Science, 333, 332–335, https://doi.org/10.1126/science.1201570.
Rajagopal, S. and A. A. Harpold, 2016: Testing and improving temperature thresholds for snow and rain prediction in the western United States. J. Amer. Water Res. Assoc., 52, 1142–1154, https://doi.org/10.1111/1752-1688.12443.
Ralph, F. M., P. J. Neiman, and G. A. Wick, 2004: Satellite and CALJET aircraft observations of atmospheric rivers over the eastern North Pacific Ocean during the El Niño winter of 1997/98. Mon. Wea. Rev., 132, 1721–1745, https://doi.org/10.1175/1520-0493(2004)132<1721:SACAOO>2.0.CO;2.
Safeeq, M., S. Shukla, I. Arismendi, G. E. Grant, S. L. Lewis, and A. Nolin, 2016: Influence of winter season climate variability on snow–precipitation ratio in the western United States. Int. J. Climatol., 36, 3175–3190, https://doi.org/10.1002/joc.4545.
Sproles, E. A., T. R. Roth, and A. W. Nolin, 2017: Future snow? A spatial-probabilistic assessment of the extraordinarily low snowpacks of 2014 and 2015 in the Oregon Cascades. Cryosphere, 11, 331–341, https://doi.org/10.5194/tc-11-331-2017.
Underwood, S. J., M. L. Kaplan, and K. C. King, 2009: The role of upstream midtropospheric circulations in the Sierra Nevada enabling leeside (spillover) precipitation. Part I: A synoptic-scale analysis of spillover precipitation and flooding in a leeside basin. J. Hydrometeor., 10, 1309–1326, https://doi.org/10.1175/2009JHM1105.1.
Ware, R. H., and Coauthors, 2000: SuomiNet: A real-time national GPS network for atmospheric research and education. Bull. Amer. Meteor. Soc., 81, 677–694, https://doi.org/10.1175/1520-0477(2000)081<0677:SARNGN>2.3.CO;2.
White, A., D. Gottas, A. Henkel, P. Neiman, F. Ralph, and S. Gutman, 2010: Developing a performance measure for snow-level forecasts. J. Hydrometeor., 11, 739–753, https://doi.org/10.1175/2009JHM1181.1.
