Daily observations of SWE and accumulated WY precipitation (the WY begins on 1 October and ends on 30 September, with the year corresponding to the latter date) were acquired from seven SNOTEL stations from the Natural Resources Conservation Survey (https://www.wcc.nrcs.usda.gov/snow/; last access: 1 June 2021) across the wUS (Fig. ; Table ). SNOTEL is a long-term, quality-controlled, surface-based network for observing precipitation and snow in wUS mountains . We used SNOTEL stations located in California, Colorado, Nevada, Utah, and Washington (Table ). These stations capture three primary snow climates (maritime, intermountain, and continental; ). We acquired daily SNOTEL data spanning the period of record (typically beginning in the 1980s) through 31 May 2020 for complete WYs. We recommend only using stations with at least 20 years of data. In our example highlighting the web-based tool, we used an end date of 8 March 2021 to show the real-time application of phase diagrams. Last, we highlight an example of how snow drought conditions influence streamflow to link phase diagrams with hydrologic outcomes. To do so, we acquired daily streamflow for WYs 1943–2019 from the US Geological Survey Gage 12082500, located on the unimpaired Nisqually River, near the Paradise, Washington, SNOTEL site (Fig. b).

To add a spatial component to station-based SWE and precipitation phase diagrams, we utilized daily gridded 4 km resolution estimates of SWE for the continental US produced by the University of Arizona (hereafter UAswe; ). The UAswe product spans WYs 1982–2020. Daily gridded 4 km spatial resolution precipitation was acquired from gridMET . Phase diagrams can be applied to any long-term daily in situ and/or gridded SWE product and are not limited to the observational products chosen in this study. Examples of watershed-averaged phase diagrams are presented and compared with nearby SNOTEL stations for two eight-digit US Geological Survey hydrologic unit code (HUC-8) watersheds in the Sierra Nevada (the upper Yuba River basin and the Tuolumne River basin).

The concept of phase diagrams was initially developed by Ludwig Boltzmann, Henri Poincaré, and Josiah Willard Gibbs . The intent was to represent all possible states of a dynamical system, such as a particle's position and momentum. Many disciplines now use phase diagrams (also referred to as phase space diagrams) – including nonlinear dynamics, chaos theory, and statistical and quantum mechanics. Each parameter of the system in phase diagrams is represented by an axis of a multidimensional space. In a two-dimensional system, each point on the phase plane represents a combination of the system's parameters, with the evolution of the system's state through time tracing a line called the phase space trajectory. The phase space trajectory begins at the point representing the initial conditions. Depending on the application, the trajectory continues indefinitely or until the time period of interest has elapsed.

Inspired by the simplicity of phase diagrams, specifically the Wheeler–Hendon phase diagrams used to track the amplitude and life cycle of the tropical intraseasonal Madden–Julian Oscillation , our purpose is to show how this visualization approach can track SWE and precipitation conditions during the cool season. We aim to track the phase space of cool-season mountain hydroclimate conditions in order to link them to the trajectory of snow drought conditions (dry and warm; ) and to the hydrometeorological events, such as rain on snow, prolonged dry spells, or periods of above-average temperature shaping the trajectories. Thus, phase diagrams can help diagnose snow drought onset, termination, duration, type, and severity as well as explore timing and characteristics of large `drought-busting” storms . By implicitly including the time-varying effects of precipitation, radiation, and temperature, phase diagrams provide a unique perspective over time series plots in tracking snowpack conditions. For instance, phase diagrams can clearly show the abrupt changes in one or both variables during major accumulation or melt events.

For each 4 km SWE grid cell, we calculated daily percentiles of median SWE from 1 October–31 May for WYs 1982–2020, again using a 7 d moving window. The same approach was performed for gridMET precipitation. Snow drought is defined similarly to above, when SWE is at or below the 30th percentile. Several boundary regions (e.g., HUC-8) were used to calculate basin averages (means) for the gridded products whose grid points fell on or within the boundary.

Cumulative discharge at the Nisqually River US Geological Survey stream gage was calculated for all complete WYs starting on the first day (1 October). For each day until the end of the WY (30 September), the accumulated discharge was calculated. For each WY, we calculated the date when 50 % of the WY total accumulated discharge occurred. Median dates of 50 % of WY total discharge were calculated using the full period of record.

WY 2020 was characterized by notable snowpack and precipitation variability throughout the cool season in the northern Sierra Nevada (Fig. a). Both fall and late winter featured near-record-low precipitation and snowpack at the Central Sierra Snow Laboratory (CSS Lab). The upper right quadrant represents wet and snowy “big year” conditions, when both accumulated precipitation and SWE exceed the 50th percentile. The upper left quadrant indicates that SWE was above the median, but accumulated precipitation was below median. Trajectories into this “dry but snowy” quadrant can result from dry fall conditions followed by appreciable snowfall, especially in places that normally receive fall precipitation as rain or in warmer lower-elevation locations when anomalous snowfall has occurred instead of rainfall. A drying fall has been identified as one signal of climatic change in California and elsewhere in the wUS . This drying could induce a systematic leftward shift in future phase diagram trajectories during the 21st century. In addition, a shift in precipitation timing into the colder months of the season could also drive a leftward shift .

Dry-snow-drought conditions (meteorological drought) are identified in the lower left (third quadrant) when SWE falls into the D0–D4 range (i.e., less than or equal to the 30th percentile), and accumulated precipitation is below the median. We defined warm snow drought when SWE falls to below the 30th percentile, and accumulated precipitation is greater than the median (lower right or fourth quadrant). An additional case of “dry but warm” snow drought occurs when trajectories are in the dry quadrant but skew towards the wetter side of a 1 : 1 line (dashed line in the “dry snow drought” quadrant). These conditions indicate drier-than-average conditions overall but with precipitation events that did not favor snowpack accumulation. To facilitate connecting various trajectories of phase diagrams with driving processes, the annotated figure is paired with a conceptual diagram showing potential physical interpretations (Fig. b).

The start of WY 2020 was characterized by bottom-third-percentile precipitation conditions with low (bottom 20th percentile) snowpack at the CSS Lab (Fig. a). Precipitation falling as snow led to rapid improvement from snow drought into the “dry but snowy” quadrant during late November into December, with precipitation recovering to near-median by mid-December. Persistent dry conditions lasting from late December through mid-March, driven by a blocking ridge west of North America , yielded snow drought onset in late January. Above-normal temperatures and dry conditions in late February and early March caused snowpack declines to accelerate, reaching a minimum value in the fifth percentile. Given that California receives the majority of its annual precipitation between December and March, dry spells lead to declines in precipitation percentile (trajectories move leftward; Fig. a). WY 2020, like other Sierra Nevada drought years, was notable for its lack of atmospheric river (AR) landfalls . ARs produce abrupt upwards and/or rightwards trajectories in the phase diagram via heavy precipitation enhanced by orography . Snow drought amelioration in late March occurred when heavy snowfall resulted from a slow-moving cutoff low-pressure system . By 1 April, the historically assumed peak timing of snowpack in the wUS (e.g., ), snow drought conditions remained, but SWE percentiles increased from the 5th to nearly the 30th percentile, though precipitation remained in the bottom 15th percentile. Another cutoff low in early April provided additional snow that terminated snow drought conditions; however accumulated precipitation remained below the median. This highlights the importance of late-spring (i.e., post-1 April) weather events in improving hydroclimatic conditions and a potential pitfall of assessing drought conditions at a single point in time. By annotating the phase diagram, the story of the cool season can be expressed to show the key events producing observed outcomes (Fig. b).

Weather events drive elevation-dependent changes in snowpack and snow drought conditions . In regions located near climatologically expected rain–snow transition elevations , such as the Sierra Nevada, individual storms can produce dramatically different responses in snowpack spatial variability and magnitude. ARs are a common type of storm event yielding variable snowpack and hydrologic responses as a result of heavy precipitation with high snowline elevations or with snowline elevations that vary widely over the duration of the storm .

WY 2018 was emblematic of the aforementioned variation in rain and snow transition elevations as both elevation-dependent and spatially dependent responses to storms and dry spells occurred in the Sierra Nevada (Fig. ). WY 2018 began with varying precipitation and SWE percentiles between three stations, again in the “dry but snowy” quadrant at the lower-elevation stations (CSS Lab and Tahoe City Cross) and near climatology for the high-elevation station (Mount Rose Ski Area). A late-November AR event was followed by a multi-month dry spell that terminated in late February. Snowpack and precipitation increased markedly in March (colloquially termed a “Miracle March”) due to persistent stormy conditions resulting from multiple landfalling ARs associated with strong midlatitude cyclones.

To highlight the elevation-induced heterogeneity of snowpack response within WY 2018, we investigate three different stations situated along a similar longitude (Fig. c). The late-November warm and wet storm caused the CSS Lab and Tahoe City Cross (Fig. a–b); both maritime snow climates) to shift rightwards and then downwards into the “warm snow drought” quadrant because much of the precipitation fell as rain. The CSS Lab is located along the Sierra Nevada crest, while Tahoe City Cross is located further east in the rain shadow of the Sierra Nevada. Unlike the other two stations, the higher-elevation Mount Rose Ski Area (hereafter “Mount Rose”), located further east in the Carson Range in a more intermountain snow climate (colder and drier than a maritime snow climate), received all snow. Mount Rose began the meteorological winter with 80th-percentile precipitation and SWE (“big year”; Fig. c). The CSS Lab and Tahoe City Cross received snow early in December, briefly moving each location out of warm snow drought. During the subsequent dry spell, the lower-elevation CSS Lab and Tahoe City Cross stations both moved leftward from warm snow drought into dry snow drought, with a 30-percentile-point decline in SWE through December. Dry-snow-drought conditions began at Mount Rose in early February. Importantly, the role of elevation is highlighted (∼ 600 m range between stations), with the colder Mount Rose experiencing less dramatic snowpack declines (reaching a minimum of the 16th percentile) compared to the warmer CSS Lab (minimum of 1st percentile) and Tahoe City Cross (minimum of 2nd percentile).

The return of North Pacific storminess during late February into March improved precipitation and snowpack conditions. This period also highlighted how snow drought amelioration is influenced by elevation. During this period, Mount Rose received all precipitation as snow. As a result, SWE increased by 30 percentile points (terminating snow drought), while precipitation increased from the 26th percentile to 52nd percentile (Fig. c). The CSS Lab increased its SWE by 35 percentile points from the lowest on record for the date in late February to non-snow-drought conditions by late March. Precipitation also increased by approximately 35 percentile points to near-median values. The cold March storms demonstrated a weaker rain shadow and generally lower snowline elevations. This increased SWE at the Tahoe City Cross from the 2nd percentile to above the 40th percentile, while precipitation also increased from the 26th to the 60th percentile between late February and early April (Fig. b). Owing to “Miracle March”, 1 April 2018 SWE conditions were closer to median than reflected by the majority of the winter, similar to WY 2020 (Fig. a). The record to near-record low late-winter SWEs at the lower-elevation CSS Lab and Tahoe City Cross are thus hidden by a single point-in-time perspective. WY 2018 and WY 2020 demonstrate the importance of a complete WY perspective regarding the assessment of evolving snow drought conditions, namely the importance of a few large precipitation events.

The warming-induced shift in precipitation phase from snow to rain has been shown in historical trends in the wUS and is projected to continue in a warmer world . Precipitation phase transition from snow to rain will result in more frequent warm snow droughts . This increase will disproportionately impact climatologically warmer maritime snow climates with important implications for the headwater hydrology and downstream reservoir management strategies of these watersheds . To provide a comparison of WYs that experience comparable snow drought conditions with respect to snowpack percentiles but differing drivers, we compared the WY 2001 dry snow drought (Fig. a) to the WY 2015 warm snow drought (Fig. b) at Paradise, Washington. Paradise is located in the Pacific Northwest on the south flank of Ti’Swaq’ (Mount Rainier; Fig. b). The WY 2015 warm snow drought in the Pacific Northwest was a motivating and formative WY for the development of snow drought research .

Paradise spent the majority of the cool season of WY 2001 in the bottom 10th precipitation percentile, a substantial difference from WY 2015, when precipitation was between the 60th and 88th percentile between December and April. The warm snow drought resulted from an anomalous amount of precipitation largely falling as rain in the early portion of winter. Snowpack conditions marginally increased throughout WY 2001 from below the 10th percentile in February to the 20th percentile by the end of the cool season (Fig. a). However in WY 2015, Paradise maintained fairly consistent SWE percentiles below the 10th percentile from February to May. The leftward trajectory of precipitation during February 2015 is indicative of drier-than-normal conditions followed by generally dry conditions (Fig. b). Weak snow drought amelioration occurred in 2001, minimizing its water resource impacts, whereas none occurred in 2015, further highlighting the importance of monitoring snow drought conditions and type over an entire WY.

Figure a–b show the entire WY phase diagrams and a SWE spatial extent snapshot at various times (Fig. d–k), relative to median climatology, for WY 2001 and WY 2015. We also highlight the differences in hydrologic outcomes between these dry- and warm-snow-drought years (Fig. c). WY 2001 had the second-lowest cumulative flows for the Nisqually River in the period studied (WY 1943–2019), but 50 % of the cumulative WY 2001 flow occurred 30 d later than the median date (3 April) at which half the Nisqually flow occurs. In contrast, WY 2015 demonstrated middle-of-the-range total WY flow (48th of 77 years) but achieved 50 % of the WY flow 56 d earlier than average. This indicates a key difference between the WYs. WY 2015 had less snow water stored later into the season than WY 2001, influencing summer streamflow. During both seasons, despite the vastly different precipitation regimes, spatial SWE anomalies are not markedly different during mid-December (Fig. d and h), mid-January (Fig. e and i), or late February (Fig. f and j). Consistent with lower SWE percentiles at Paradise during WY 2015 compared to WY 2001, as shown on the phase diagrams, SWE anomalies are modestly more negative. The lack of mountain snowpack during WY 2015 was more notable than WY 2001 (Figs. g and k). The comparatively better spring snowpack in WY 2001 likely helped maintain flows later into the year despite an otherwise dry year.

Enhanced interannual precipitation variability and decreased peak SWE variability are expectations of a warming climate . Comparisons of extreme years and their outcomes provide valuable object lessons for water managers and other resource planners . Phase diagrams allow for direct comparisons between WYs, helping to identify key differences in hydroclimate variability for a particular region of interest. For example, Red Mountain Pass, located in a high-elevation, continental snow climate within the San Juan Mountains of southwestern Colorado, is used to compare two late-cool-season outcomes that represent two hydroclimatic extremes. The majority of WY 2011 showed phase trajectories in the “big year” first quadrant (Fig. a) after a slightly below-average start to snowpack totals between October and early December. A stormy December increased SWE and precipitation percentiles. Stormy weather continued in April and May, preventing snowmelt and causing precipitation and SWE percentiles to increase. WY 2012 began with above-average precipitation and snowpack in fall but drier-than-normal conditions throughout winter, which resulted in dry-snow-drought onset in December (Fig. b). Modest snow drought amelioration occurred in early March, but with a few exceptions in April, dry conditions persisted through May. This led to the onset of dry snow drought, again, via rapid snowmelt and below-average precipitation.

Spatial SWE distributions are consistent with the phase diagrams (Fig. c–j). In both years, SWE anomalies increased throughout the accumulation season and then accelerated in late spring. Compared to the emerging drought signal during WY 2012, WY 2011 did not demonstrate widespread positive SWE anomalies throughout the year. Between January and April, lower-elevation regions experienced below-normal SWE anomalies (Fig. c–d), whereas higher elevations had above-normal SWE. This difference resulted from above-normal temperatures and below-normal precipitation, likely driven by snow-albedo feedbacks that were enhanced at lower elevations where snowmelt occurred .

We applied SNOTEL station data to create the phase diagrams, but a challenge in mountain environments is the lack of reliable, well-distributed, long-term observations. In lieu of station data, gridded observational products commonly inform natural resource decision-making and research efforts. The necessary components exist to create phase diagrams using gridded meteorological products , observation-based snow datasets , and output from hydrological simulations . The challenge is how to aggregate spatial information to extract maximum information value for the application of interest (e.g., water management, avalanche forecasting, ecosystem processes) regarding the state and evolution of snowpack conditions at relevant scales (e.g., the full watershed, a sub-basin, or within specific elevation bands) in complex terrain. We have performed a first step towards this goal. Initial methods to broaden the approach could be performed by (1) binning regions by similar elevation, watershed, slope, aspect, and/or land cover type; (2) identifying areas that co-vary together in time and space using techniques such as principal component or cluster analysis; and (3) subjective grouping based on anecdotal information from managers. Creating meaningful phase diagrams for varied management and scientific applications using spatially distributed information is the primary goal of our ongoing research. This will allow evaluation of snow drought in regions without long-term snow observing networks such as in the northeastern US or other high mountain areas worldwide. To this end, we next provide examples of how spatially distributed products can visualize snow drought.

A snow drought tracker web application has been developed to provide users with access to snow drought phase diagrams. The beta version is available at https://wrcc.dri.edu/my/climate/snow-drought-tracker and simply requires users to set up a username and password to access it. The web tool is updated in near real time across the wUS and Alaska. The SNOTEL network is the backbone of the tool with over 800 stations providing daily SWE, snow depth, precipitation, and temperature data. An interactive map (Fig. a–b) allows for station selection via a graphical user interface. Once a station is selected, current-year observations are displayed on the dashboard by default (Fig. c–d). As an example, the Mill-D North station in the Wasatch Mountains of northern Utah is shown (Fig. b). The dashboard will also display the most recent daily updated snow drought phase diagram (starting on 1 October of WY 2021; Fig. e). The Almanac has several tabs showing daily SWE, snow depth, and precipitation absolute values and percent of average SWE (Fig. c). Note that because percent of average is more commonly used by managers, the first iteration of the tool uses percent of average (and the 80 % threshold used by ) instead of percentiles and drought monitor thresholds as otherwise focused on in this paper. From the Almanac the month-to-date, calendar-year-to-date, and WY-to-date precipitation values and percent of normal can be viewed (Fig. d). In addition to the current-year data found on the dashboard, historical data and graphics can be generated. Phase diagrams can be created for any year in the station record, and daily time series plots can be generated for SWE, snow depth, precipitation, and temperature (e.g., Fig. e). Figures can be downloaded as PNG or SVG files, and historical data can be downloaded in CSV format. Beta testing of the snow drought tracker is being conducted by stakeholders including the National Weather Service, California Department of Water Resources, and state climatologists around the wUS. Other agencies will be encouraged to test the tool after the first round of testing and updates have concluded. Feedback from the testing will be incorporated into future upgrades of the snow drought tracker, with the goal of further developing a web-based product that provides a science-to-service-to-practice interface . Some known limitations and gaps in the current version of the tool include the lack of spatial snow drought information (i.e., the river basin composites; Figs. , , and created using gridded products) and the need to incorporate elevation gradients into snow drought monitoring.