Abstract Seasonal forecasting in operational centers has focused predominantly on prediction of temperature and precipitation. Here, we use the skill of model forecasts against observations (actual prediction skill) and against each model ensemble member (perfect-model skill) to assess the predictability of these two variables and five additional hydrological variables across two distinct hydrologic regions (the Missouri basin and California) of the United States. The forecasts from two operational coupled models [National Centers for Environmental Prediction Climate Forecast System version 2 (NCEP CFSv2) and European Centre for Medium-Range Weather Forecasts Seasonal Forecasting System version 5 (ECMWF SEAS5)] are used. Results show variables with high memory, such as soil moisture, total water storage, and snow water equivalent, have both high perfect-model and actual prediction skill. Runoff and evapotranspiration, which are highly dependent on the amount of water storage, generally have intermediate skill. Subbasins within these two hydrologic regions show similar results. The high memory variables also maintain high skill for longer prediction lead and exhibit less seasonal variability, particularly over the Missouri basin region which experiences a lower amplitude in the annual cycle. However, over California, skill drops off more quickly with increasing lead and has a strong seasonal cycle with the highest skill during late spring and early summer. Results also demonstrate that actual prediction skill is positively related to perfect-model skill for all variables across different regions and seasons which give some promise to using perfect-model skill in ungauged areas of the world as a proxy to their real-world skill. Significance Statement Operational seasonal forecasting models are commonly evaluated based on their predictions of temperature and precipitation. Here, we assess the forecasts for these two variables and five additional variables associated with the hydrologic cycle across two distinct hydrologic regions of the United States. We find that total water storage, soil moisture, and snowpack are more dependent on past values (i.e., with high memory) and hence have the highest predictability. Evapotranspiration and runoff, which are highly dependent on water storage, have higher predictability than temperature and precipitation. Also, the variables and regions with the highest memory maintain higher skill at longer lead times and have consistent skill throughout the year. Basins and seasons that have higher intrinsic model predictability also have higher skill predicting real-world values for all variables.

In arid and semi-arid regions, groundwater sustains vegetation through subsurface water access, yet the responses of groundwater-dependent ecosystems (GDEs) to changing hydroclimate and groundwater availability are relatively understudied. This study investigates seasonal and spatial patterns in vegetation greenness using Landsat Enhanced Vegetation Index (EVI) values across riparian and upland zones in the semi-arid Upper San Pedro (USP) watershed, southern Arizona, which experiences a bimodal precipitation regime. We paired 25 years (2000-2024) of EVI and depth to groundwater (DTG) data from 89 wells and climate metrics (precipitation and vapour pressure deficit) to quantify the sensitivity of vegetation to subsurface moisture as well as atmospheric moisture supply and demand. Vegetation at wells near the USP riparian area showed strong associations between EVI and DTG anomalies during the monsoon season, indicating sustained groundwater use even during this wet period when summer precipitation is abundant. In contrast, upland vegetation that lacked access to groundwater showed minimal sensitivity in EVI to DTG and was generally less responsive to vapour pressure deficit. Interestingly, the riparian GDEs were not decoupled from precipitation and climate variability. These results underscore the importance of groundwater for maintaining riparian productivity and highlight the utility of remote sensing in identifying vegetation-climate-groundwater linkages across heterogeneous dryland landscapes.

Forests are changing rapidly due to drought, disease, wildfire, and forest management, with unknown impacts on snowmelt resources. While hyper-resolution forest hydrology models capture the effects of canopy cover amount and arrangement on snowpack, they are too complex for landscape-scale assessments. Here, we evaluated two statistical approaches to emulate high-resolution (1 m) mechanistic model maps of snow variables for future application in forest management and drew inferences about topographic vs. forest canopy controls. We tested a simple landscape classification using one or more of topographic northness, canopy cover, and surrounding forest arrangement and machine learning (ML) models of varying complexity to emulate maps of peak SWE, liquid water input, and snow cover duration (SCD) from the 3-D forest hydrology model SnowPALM, which was previously trained using lidar and daily Snowtography. We evaluated three winters at three mid-scale study areas (similar to 50 ha) in the southwestern US spanning gradients of SCD and forest type. All approaches emulated areal mean values within <2 %. Landscape classification captured 50-80 % of spatial variability, while ML explained 88-98 %. Increasing ML complexity was needed to emulate snow maps having greater spatial variability, which tended to occur where SCD was greatest: at high/cold sites, in cold/wet winters, and in locations shaded by terrain or nearby trees. Warm/dry forests were adequately modeled using canopy cover. These results demonstrate a generalizable approach for future upscaling of forest snow measurements through sequential mechanistic and statistical modeling for hydrologically informed forest management.

Despite having important implications for water resources, the climatic dependence of forest thinning impacts on snowpack is poorly quantified. In this study, we used a high-resolution snow model to understand the impact of forest thinning on snowpack in Arizona under contrasting climate conditions, leading to ephemeral vs. seasonal snowpack conditions. The model is evaluated using a spatiotemporally extensive set of snowpack measurements and is run for the same set of pre- and post-thinning forest patch geometry using two meteorological forcing datasets representing locally mid- and high-elevation climate conditions. Although the high-elevation climate is only 1 degrees C cooler and has 20% more winter precipitation, it leads to markedly different snowpack conditions, i.e., twice as long-lasting snowpack, less mid-winter ablation events and similar to 60% larger at its peak. For both climates, forest thinning increased peak snow water equivalent (SWE) and liquid water input (LWI), but it decreased snow cover duration (SCD) only for the high-elevation climate. Total sublimation losses decreased from similar to 35% of wintertime precipitation pre-thinning to similar to 25% post-thinning for the high-elevation climate and from similar to 25% to similar to 15% for the mid-elevation climate. Generally, a 10% reduction in canopy cover resulted in similar to 4.5% more snowfall reaching the ground, and a 10-day decrease in SCD reduced the fraction of winter precipitation lost to snowpack sublimation by similar to 2%. Post-thinning changes in forest patch geometry were also important as larger canopy gaps had more LWI, and areas with warmer canopy edges had lower peak SWE and SCD.

Montane snowpack in the Sierra Nevada provides critical water resources for ecological functions and downstream communities. Forest removal allows us to manage the snowpack in montane forests and mitigate the effect of climate on water resources. Little is known about the mid- to long-term effects that changing snowpack following forest disturbance has on tree re-growth, and how tree re-growth might in turn affect snowpack accumulation and melt. We use a 1-m resolution process-based snow model (SnowPALM) coupled with a standscale ecohydrological model (RHESSys) that resolves water, energy and carbon cycling to represent tree growth, and to quantify how trees and snowpack co-evolve following two disturbance scenarios (thinning and clearcutting) over a period of 40 years in a small 100 m x 234 m mid-elevation forested area in the Sierra Nevada, California. We first calculate the impact of forest disturbance on the snowpack assuming no tree regrowth and then we compare it with scenarios that include the feedback of trees regrowth on the snowpack. Without tree regrowth, snow accumulation and melt volume increase on average by roughly 5 % and 13 % following thinning and clearcutting, respectively. With tree regrowth, a regrowth rate of 0.75 and 1.15 m/decade are found for thinning and clearcutting, respectively, along with a decrease of melt volumes of 2.5 to 0.9 mm/decade, respectively. About 50 % of the snowmelt volume gains from forest thinning are lost after 40 years of regrowth, whereas only about 7 % is lost from clearcutting after the same period, which are largely explained by changes to canopy interception and sublimation. This proof-of-concept study is expected to shed light into the coevolution of montane forests and snowpack response to forest disturbance.

California experienced a historic run of nine consecutive landfalling atmospheric rivers (ARs) in three weeks' time during winter 2022/23. Following three years of drought from 2020 to 2022, intense landfalling ARs across California in December 2022-January 2023 were responsible for bringing reservoirs back to historical averages and producing damaging floods and debris flows. In recent years, the Center for Western Weather and Water Extremes and collaborating institutions have developed and routinely provided to end users peer-reviewed experimental seasonal (1-6 month lead time) and subseasonal (2-6 week lead time) prediction tools for western U.S. ARs, circulation regimes, and precipitation. Here, we evaluate the performance of experimental seasonal precipitation forecasts for winter 2022/23, along with experimental subseasonal AR activity and circulation forecasts during the December 2022 regime shift from dry conditions to persistent troughing and record AR-driven wetness over the western United States. Experimental seasonal precipitation forecasts were too dry across Southern California (likely due to their overreliance on La Nina), and the observed above-normal precipitation across Northern and Central California was underpredicted. However, experimental subseasonal forecasts skillfully captured the regime shift from dry to wet conditions in late December 2022 at 2-3 week lead time. During this time, an active MJO shift from phases 4 and 5 to 6 and 7 occurred, which historically tilts the odds toward increased AR activity over California. New experimental seasonal and subseasonal synthesis forecast products, designed to aggregate information across institutions and methods, are introduced in the context of this historic winter to provide situational awareness guidance to western U.S. water managers.

Lateral subsurface flow plays an essential role in sustaining the terrestrial ecosystem, but it is not explicitly represented in most Earth System Models. In this study, we implemented an explicit lateral saturated flow model into the E3SM land model (ELM). The model explicitly describes lateral flow in the saturated zone by representing, for each model grid, an idealized hillslope consisting of five hydrologically connected soil columns. We conducted three model experiments driven by 0.125 degrees atmospheric forcing data during 1980-2015 over California using models of the default ELM, a modified version of ELM to enhance infiltration, and the model with the lateral saturated flow model. The simulated runoff, evapotranspiration, and terrestrial water storage anomaly (TWSA) from the three simulations were evaluated against available observations, and the model explicitly representing lateral flow performs best. The new model produces greater gridcell-averaged evapotranspiration especially over the mountainous regions with moderate relief and seasonally dry climates. Most importantly, it improves the modeled seasonal variations, interannual variabilities, and the recent decadal decline of TWSA. Many of these improvements can be attributed to the enhanced ecosystem resilience to droughts as demonstrated by transpiration increases caused by lateral flow. Model sensitivity experiments suggest that subsurface runoff is most sensitive to the ratio between horizontal and vertical saturated hydraulic conductivity, followed by hillslope planforms (convergent, divergent, and uniform), number of columns, and lower boundary conditions. Future work should effectively characterize hillslopes in global models and explore the long-term influences of lateral water movement on modeled biogeochemical cycle.

Across the western United States, forests are changing rapidly, with uncertain impacts on snowmelt water resources. Snow partitioning is controlled by forest effects on interception, radiation, and sublimation. Yet, models often lack snow measurements with sufficiently high spatial and temporal resolution across gradients of forest structure to accurately represent these fine-scale processes. Here, we utilize four Snowtography stations in Arizona, in the lower Colorado River Basin, with daily measurements over 3–5 years at ~110 positions distributed across gradients of forest structure resulting from wildfires and mechanical thinning. We combine Snowtography with lidar snapshots of forest and snow to train a high-resolution snow model and run it for 6 years to quantify how forest structure regulates snowpack and snowmelt. These study sites represent a climate gradient from lower/warmer ephemeral snowpack (~2,100 m asl) to higher/colder seasonal snowpack (~2,800 m asl). Forest cover reduced snowpack and snowmelt through canopy sublimation. Forest advanced snowmelt timing at lower/warmer sites but delayed it at higher/colder sites. Within canopy gaps, shaded cool edges had the greatest peak snow water equivalent (SWE). Surprisingly, sunny/warm gap edges produced more snowmelt than cool edges, because high radiation melted snow quickly, reducing exposure to sublimation. Therefore, peak SWE is not an ideal proxy for snowmelt volume from ephemeral snowpacks, which are becoming more prevalent due to warming. The results imply that forest management can influence the amount and timing of snowmelt, and that there may be decision trade-offs between enhancing forest resilience through delayed snowmelt and maximizing snowmelt volumes for downstream water resources.

Abstract Accurate mapping of snow amount in the mountains is critical as mountain snowpacks are water supply for millions of people. Satellite remote sensing has been largely unable to reliably detect the amount of snowpack in these areas. Recently, C‐band Synthetic Aperture Radar (SAR) data from the Sentinel‐1 (S1) satellites have shown potential for measuring snow depth in the mountains. However, their spatiotemporal coverage is incomplete, and their evaluation with robust, aerial snow depth data is limited. Here, we evaluate two S1 snowpack datasets with some of the best available gridded snowpack data over the Colorado Rockies and Sierra Nevada mountains in the western US: the Airborne Snow Observatory (ASO) and the University of Arizona (UA) snowpack datasets. Compared to ASO and UA data, the S1 data are biased high when snow is shallow, and biased low when snow is deep (particularly later in spring when there is wet snow), though these biases are reduced for deep snow areas when wet snow pixels are removed. We then apply corrections based on machine learning that account for physiographic characteristics to improve the accuracy of the S1 data. Furthermore, we fill gaps in the S1 data by using snow persistence, but also account for potential snow accumulation and ablation, to generate temporally complete snow depth maps over mountainous areas. Corrected and gap‐filled S1 snow depth mapping could be especially important for snow monitoring in remote mountain areas where other techniques for snow mapping do not work or are logistically infeasible or cost‐prohibitive.

In the western US, most rangelands receive snowfall. Yet, a commonly used tool to assess rangeland's vulnerability to erosion, the USDA's Rangeland Hydrology and Erosion Model (RHEM) is run using long-term simulated climate inputs that assumes that all precipitation occurs as rainfall. This can be problematic for areas that receive heavy snowfall or substantial rain-on-snow events. In this research, we have developed an efficient snow module for RHEM, called RHEM-Snow, which partitions precipitation between rainfall and snowfall, simulates snowpack accumulation and ablation, and passes net water input (consisting of rainfall, snowmelt, or both) to RHEM. In some areas, the inclusion of the snow module can reduce annual overland flow runoff and erosion estimates by more than 20 % of the total annual overland flow runoff and erosion produced without the snow module (or by as much as 10-50 mm/year for overland flow runoff or >100 kg/ha-yr for erosion). The reclassification of precipitation events from rainfall in RHEM to snowfall in RHEM-Snow tends to reduce overland flow runoff and erosion, but this reduction can be partially counterbalanced by increases from snowmelt and rain-on-snow. However, hydrologic responses to rain-on-snow events can either be enhanced or muted depending on the characteristics of the storm and the snowpack, as sometimes the snowpack can absorb the precipitation inputs, and sometimes snowmelt enhances the precipitation inputs. Because of this mixed impact, the average difference in erosion caused by rain on snow events is relatively small compared to corresponding events where only the liquid phase is considered. Further study is needed of the complex erosion processes under snowpack and frozen soil/variable saturation conditions. Overall, RHEM-Snow provides more realistic timing and magnitude of overland flow runoff and erosion in cold environments, better satisfying the conditions for RHEM applications.

Abstract Snowfall forecasting has historically been an area of difficulty for operational meteorologists, particularly in regions of complex terrain, such as the western United States. Attempts at improving forecasts have been made, but skill is still poor, with snowfall routinely overpredicted. A major reason for this overprediction has been the failure to accurately predict snow–liquid ratios (SLR) ahead of major events. This research proposes, develops, and tests multiple machine learning methods for dynamic SLR prediction for the Sky Islands of southeast Arizona by objectively comparing a multiple linear regression (MLR) against several more complex and flexible machine learning methods. Input parameters for each method were chosen based on variables found by previous studies to have a regression‐based relationship with SLR, with a focus on the lower mid‐levels of the troposphere. These parameters were also used to construct the MLR model, and its performance was compared objectively with the machine learning methods. When tested on historical events, a very high percentage of the network‐predicted SLR values fall within the margin of error of observed SLRs, which were calculated using gridded snow depth and snow water equivalent (SWE) data from the University of Arizona daily 4‐km SWE, SD, and SCE dataset (UASnow). A support vector machine (SVM), a k‐nearest neighbor (KNN) algorithm, and a random forest also showed high accuracies when tested on the dataset, and each showed a significant gain in skill compared with the MLR model, with skill being evaluated by multiple metrics.

Forested, snow-dominated watersheds provide a range of ecosystem services including water supply, carbon sequestration, habitat and recreation. While hydrologic partitioning has been well-studied in watersheds with stable seasonal snowpack, less is known about watersheds with ephemeral snowpack. Furthermore, drought-related disturbances and/or management practices are altering vegetation cover in many forests, with unknown and potentially different, consequences for stable seasonal versus ephemeral snowpacks. This study quantifies net water input (NWI) to soil for two sites with contrasting stable seasonal and ephemeral snowpacks, respectively, for three water years in Arizona, USA. Observations include a network of automated cameras and graduated snow stakes (snowtography) deployed across gradients of forest structure, airborne lidar maps of topography and forests and SNOTEL station records. Given the importance of mixed-phase precipitation in ephemeral snowpack watersheds, an algorithm is developed to distinguish among snowfall and rainfall that does/does not contribute to snowpack mass. Finally, existing canopy interception and snowpack models are used to estimate how NWI varies with canopy cover. At the ephemeral snowpack site, increasing canopy cover reduces NWI amount and advances its seasonal timing less strongly than at the stable seasonal snowpack site. Interestingly, canopy reduces NWI duration at the ephemeral site but prolongs it at the stable seasonal snowpack site. These effects are more important in a cool/wet and average year than a warm/dry year. Understanding differences between canopy impacts on amount, timing and duration of NWI for areas with ephemeral versus stable seasonal snowpack is increasingly important as the number of watersheds with ephemeral snowpack grows.

Forested, snow-dominated watersheds provide a range of ecosystem services for adjacent and downstream communities including water supply, wildlife habitat, and recreation. While forest-snow-water relations have been well studied in areas with stable seasonal snowpacks, less is known about these interactions in the growing number of areas with ephemeral snowpacks. Here, we evaluate how forest cover and climate variability regulate the seasonality of water inputs, soil moisture, root zone water stress, and percolation at two contrasting seasonal vs. ephemeral snowpack sites, in Arizona, USA. We use a soil moisture model (Hydrus-1D) and three years of continuous soil moisture measurements at two depths (0-10 cm and 0-30 cm) for 6 locations/site distributed across gradients of forest cover. Each soil moisture measurement location has a unique daily moisture input time series previously developed from local snow/rain observations and modeling. Compared to the seasonal snowpack site, the ephemeral site has wetter soils during winters due to frequent mid-winter melt events, leading to greater annual percolation. However, earlier snowmelt at the ephemeral snowpack site also prolongs the duration of root zone water stress. At both sites, dense forest cover is associated with less percolation and longer duration of soil drought, due to increased transpiration. These effects are amplified during an abnormally dry winter. This study examines two sites over three years and develops key steps towards using spatially distributed snowpack and soil moisture measurements to assess how changing climate and forest cover jointly regulate water availability for future ecosystems and downstream use.

The Sierra Nevada has experienced unprecedented wildfires and reduced snowmelt runoff in recent decades, due partially to anthropogenic climate change and over a century of fire suppression. To address these challenges, public land agencies are planning forest restoration treatments, which have the potential to both increase water availability and reduce the likelihood of uncontrollable wildfires. However, the impact of forest restoration on snowpack is site specific and not well understood across gradients of climate and topography. To improve our understanding of how forest restoration might impact snowpack across diverse conditions in the central Sierra Nevada, we run the high-resolution (1 m) energy and mass balance Snow Physics and Lidar Mapping (SnowPALM) model across five 23-75 km2 subdomains in the region where forest thinning is planned or recently completed. We conduct two virtual thinning experiments by removing all trees shorter than 10 or 20 m tall and rerunning SnowPALM to calculate the change in meltwater input. Our results indicate heterogeneous responses to thinning due to differences in climate and wind across our five central Sierra Nevada subdomains. We also predict the largest increases in snow retention when thinning forests with tall (7-20 m) and dense (40-70% canopy cover) trees, highlighting the importance of pre-thinning vegetation structure. We develop a decision support tool using a random forests model to determine which regions would most benefit from thinning. In many locations, we expect major forest restoration to increase snow accumulation, while other areas with short and sparse canopies, as well as sunny and windy climates, are more likely to see decreased snowpack following thinning. Our decision support tool provides stand-scale (30 m) information to land managers across the central Sierra Nevada region to best take advantage of climate and existing forest structure to obtain the greatest snowpack benefits from forest restoration.

Operational streamflow forecasting is critically important to managers of river basins that supply water, hydropower, and flood protection. While seasonal water supply forecasts (WSFs) are important for long-term water resources planning operations, shorter term (e.g., 1-5 weeks) streamflow forecasts are critical for balancing water conservation with flood risk during wet periods. In this study, we designed a streamflow forecasting system with the water resources group at the Salt River Project (SRP), a provider of water and power to millions of customers in central Arizona (AZ), to provide streamflow forecasts for a diverse and operationally important set of watersheds in AZ. The forecast system uses machine learning to make seasonal WSFs, a rainfall-runoff model driven by ensemble meteorological forecasts to make 35-day streamflow forecasts, and an innovative approach to improve the WSFs based on the 35-day streamflow forecasts. This model integration allows for an assessment of the impact of different meteorological forecasts on WSFs, helping SRP to balance water conservation goals with shorter term flood risks. In addition, seasonal WSFs are improved in the early winter when they incorporate the 35-day streamflow predictions. Furthermore, these improvements are larger than when they incorporate 7-day streamflow predictions, demonstrating the value of using subseasonal to seasonal (S2S, >1-2 weeks) forecasts to improve seasonal WSFs in these watersheds.

Accurately modeling the effects of variable forest structure and change on snow distribution and persistence is critical to water resource management. The resolution of many snow models is too coarse to represent heterogeneous canopy structure in forests, and therefore, most models simplify forest effects on snowpack mass and energy budgets. To quantify the loss of snowpack prediction from simplifications of forest canopy‐mediated processes, we applied a high‐resolution energy balance snowpack model at two forested sites at a fine (1 m 2 ) and coarse (100 m 2 ) spatial resolution. Simulating open and forested areas separately, as is done in many land surface models (LSMs), leads to biases between the coarse and fine‐scale simulations because there is no representation of areas that are near (e.g., <15 m from) trees but with no overhead canopy, which are common in forests of low to medium tree density. Consistent with previous LSM intercomparisons, the coarser simulations predict greater under‐canopy radiation (by 30%–80% at our sites), faster snow ablation (by almost 2×), and earlier snow disappearance (by 1–22 days). Many of these biases are reduced dramatically or eliminated when canopy edge environments are considered in the coarser simulations. Furthermore, remaining disagreement between the 100‐m and 1‐m models can be partially explained by using a combination of tree height, canopy cover, and canopy edginess (which together can explain 46%–96% of remaining model biases). The lack of information about canopy edges and other fine‐scale forest structure characteristics in many current LSMs may limit their reliability for simulating forest disturbance.

The combination of snowfall, snow water equivalent (SWE), and precipitation rate measurements from 39 snow telemetry (SNOTEL) sites in Alaska were used to assess the performance of various precipitation products from satellites, reanalysis, and rain gauges. Observation of precipitation from two water years (2018–2019) of a high-resolution radar/rain gauge data (Stage IV) product was also utilized to give insights into the scaling differences between various products. The outcomes were used to assess two popular methods for rain gauge undercatch correction. It was found that SWE and precipitation measurements at SNOTELs, as well as precipitation estimates based on Stage IV data, are generally consistent and can provide a range within which other products can be assessed. The time-series of snowfall and SWE accumulation suggests that most of the products can capture snowfall events; however, differences exist in their accumulation. Reanalysis products tended to overestimate snow accumulation in the study area, while the current combined passive microwave remote sensing products (i.e., IMERG-HQ) underestimate snowfall accumulation. We found that correction factors applied to rain gauges are effective for improving their undercatch, especially for snowfall. However, no improvement in correlation is seen when correction factors are applied, and rainfall is still estimated better than snowfall. Even though IMERG-HQ has less skill for capturing snowfall than rainfall, analysis using Taylor plots showed that the combined microwave product does have skill for capturing the geographical distribution of snowfall and precipitation accumulation; therefore, bias adjustment might lead to reasonable precipitation estimates. This study demonstrates that other snow properties (e.g., SWE accumulation at the SNOTEL sites) can complement precipitation data to estimate snowfall. In the future, gridded SWE and snow depth data from GlobSnow and Sentinel-1 can be used to assess snowfall and its distribution over broader regions.

Snowmelt from mountain forests is critically important for water resources and hydropower generation. More than 75% of surface water supply originates as snowmelt in mountainous regions, such as the western U.S. Remote sensing has the potential to measure snowpack in these areas accurately. In this research, we combine light detection and ranging (lidar) from crewed aircraft (currently, the most reliable way of measuring snow depth in mountain forests) and structure from motion (SfM) remotely piloted aircraft systems (RPAS) for cost-effective multi-temporal monitoring of snowpack in mountain forests. In sparsely forested areas, both technologies give similar snow depth maps, with a comparable agreement with ground-based snow depth observations (RMSE ~10 cm). In densely forested areas, airborne lidar is better able to represent snow depth than RPAS-SfM (RMSE ~10 cm vs ~10–20 cm). In addition, we find the relationship between RPAS-SfM and previous lidar snow depth data can be used to estimate snow depth conditions outside of relatively small RPAS-SfM monitoring plots, with RMSE’s between these observed and estimated snow depths on the order of 10–15 cm for the larger lidar coverages. This suggests that when a single airborne lidar snow survey exists, RPAS-SfM may provide useful multi-temporal snow monitoring that can estimate basin-scale snowpack, at a much lower cost than multiple airborne lidar surveys. Doing so requires a pre-existing mid-winter or peak-snowpack airborne lidar snow survey, and subsequent well-designed paired SfM and field snow surveys that accurately capture substantial snow depth variability.

In the Southwest United States, water resources depend heavily on snowpacks, which are temporally and spatially limited in this warm, semiarid region. Snow accumulation and ablation in the Southwest are heavily influenced by forest structure. Therefore, water resource managers urgently need to understand the future impacts of unprecedented forest changes now occurring from drought, insect infestation, and forest management. Here, we present state‐of‐the‐art maps and time series of snow water equivalent (SWE), which account for spatial variability of snow depth and snow density over a large range of forest structure and topographies in the highlands of central Arizona. We show that the degree of forest cover and its geometry largely determine which areas have more/less snow accumulation and faster/slower ablation. Compared with under‐canopy areas, open areas can have 20–30% more accumulation and ablation rates can vary by 15–30% in sunny areas versus shaded areas. Although SWE response to forest cover is widely variable, depending on how much shading the trees provide versus how much snow is intercepted and lost through canopy sublimation, dense forests generally have less SWE than sparser forests. In general, SWE is optimized at intermediate levels of forest cover (~30–50%) on flat and north‐facing slopes. Somewhat counterintuitively, increasing forest cover generally causes a reduction of SWE on south‐facing slopes, where trees are less effective at reducing radiative forcing at the snow surface due to less efficient shading and increased enhancement of longwave radiation from the warm canopy.