Figure 1 - available via license: Creative Commons Attribution-NonCommercial 4.0 International
Content may be subject to copyright.
The Colorado River Basin in the Southwestern United States with the location of the snow telemetry (SNOTEL) stations. Sample stations that illustrate different climatic regions are labeled and highlighted with an encircled dot. Figura 1. La Cuenca del Río Colorado en el suroeste de Estados Unidos con la ubicación de las estaciones de telemetría de nieve (SNOTEL). Se han etiquetado las estaciones de muestreo que ilustran las diferentes regiones climáticas y se destacan con un punto rodeado.
Source publication
The relation between snow water equivalent (SWE) and 28 variables (27 topographically-based topographic variables and canopy density) for the Colorado River Basin, USA was explored through a multi-variate regression. These variables include location, slope and aspect at different scales, derived variables to indicate the distance to sources of mois...
Contexts in source publication
Context 1
... study area is the Colorado River Basin in the southwestern US ( Figure 1 La Cuenca del Río Colorado en el suroeste de Estados Unidos con la ubicación de las estaciones de telemetría de nieve (SNOTEL). Se han etiquetado las estaciones de muestreo que ilustran las diferentes regiones climáticas y se destacan con un punto rodeado. ...
Citations
... Current snowpack monitoring does not represent the range of land covers, elevations and terrain classes found in mountainous regions (Fassnacht et al., 2012;Kampf et al., 2020;Sexstone & Fassnacht, 2014), partially due to the difficulties of measuring snow in alpine areas. Stations do not always reflect their nearby surroundings (Meromy et al., 2013), and conditions in mountainous regions experience substantial inter-annual variability (Fassnacht et al., 2003(Fassnacht et al., , 2012 especially in the alpine (Sexstone et al., 2018). ...
... Current snowpack monitoring does not represent the range of land covers, elevations and terrain classes found in mountainous regions (Fassnacht et al., 2012;Kampf et al., 2020;Sexstone & Fassnacht, 2014), partially due to the difficulties of measuring snow in alpine areas. Stations do not always reflect their nearby surroundings (Meromy et al., 2013), and conditions in mountainous regions experience substantial inter-annual variability (Fassnacht et al., 2003(Fassnacht et al., , 2012 especially in the alpine (Sexstone et al., 2018). The 75 existing SNOTEL sites in the study region range in elevation from 2,605 to 3,532 m, with a mean elevation of 3,094 and median of 3,127 m. ...
Abstract Continued climate warming is reducing seasonal snowpacks in the western United States, where >50% of historical water supplies were snowmelt‐derived. In the Upper Colorado River Basin, declining snow water equivalent (SWE) and altered surface water input (SWI, rainfall and snowmelt available to enter the soil) timing and magnitude affect streamflow generation and water availability. To adapt effectively to future conditions, we need to understand current spatiotemporal distributions of SWE and SWI and how they may change in future decades. We developed 100‐m SnowModel simulations for water years 2001–2013 and two scenarios: control (CTL) and pseudo‐global‐warming (PGW). The PGW fraction of precipitation falling as snow was lower relative to CTL, except for November–April at high elevations. PGW peak SWE was lower for low (−45%) and mid elevations (−14%), while the date of peak SWE was uniformly earlier in the year for all elevations (17–23 days). Currently unmonitored high elevation snow represented a greater fraction of total PGW SWE. PGW peak daily SWI was higher for all elevations (30%–42%), while the dates of SWI peaks and centroids were earlier in the year for all elevations under PGW. PGW displayed elevated winter SWI, lower summer SWI, and changes in spring SWI timing were elevation‐dependent. Although PGW peak SWI was elevated and earlier compared to CTL, SWI was more evenly distributed throughout the year for PGW. These simulated shifts in the timing and magnitude of SWE and SWI have broad implications for water management in dry, snow‐dominated regions.
... The relationships between snowpack, topography, and forest coverage have been pointed out in previous studies: among others, Erxleben et al. [28] mapped SWE by accounting for the forest coverage and topographic parameters as the aspect, the elevation, and the local slope. Fassnacht et al. [29] exploited the SWE dependence on location, aspect, slope, and canopy density. ...
This study aims at estimating the dry snow water equivalent (SWE) by using X-band synthetic aperture radar (SAR) data from the COSMO-SkyMed (CSK) satellite constellation. The time series of CSK acquisitions have been collected during the dry snow period in the Alto Adige test site, in the Italian Alps, during the winter seasons from 2013 to 2015 and from 2019 to 2021. The SAR data have been analyzed and compared with the
in situ
measurements to understand the X-band SAR sensitivity to SWE, which has been further assessed by dense media radiative transfer (DMRT) model simulations. The sensitivity analysis provided the basis for addressing the SWE retrieval from the CSK data, by exploiting two different machine learning (ML) techniques, namely, artificial neural networks (ANNs) and support vector regression (SVR). To ensure statistical independence of training and validation processes, the algorithms are trained and tested using SWE predictions of the fully distributed snow model AMUNDSEN as reference data and are subsequently validated on the experimental dataset. Due to its influence on the CSK estimates, the effect of forest canopy was accounted for in the analysis. Depending on the algorithm, the validation resulted in a correlation coefficient
$0.78\le R \le0.91$
and a root-mean-square error (RMSE) 55.5 mm
$\le $
RMSE
$\le87.4$
mm between estimated and
in situ
SWE. Further analysis and validation are needed; however, the obtained results seem suggesting the CSK constellation as an effective tool for the retrieval of the dry SWE in alpine areas.
... They collect and analyze snow data from snow courses, and now there are more than 200 snow courses located throughout the Sierra Nevada. These ground-based SWE measurements have been interpolated into spatially distributed SWE based on their relationships with predictor variables (e.g., elevation, aspect, slope, solar radiation, wind, etc.) using spatial interpolation or regression methods (Carroll et al., 1999;Fassnacht et al., 2003;Fassnacht et al., 2012;. However, the accuracy of these statistcical methods is largly hindered by the high heterogeneity of SWE distribution over mountainous terrian, the observation desnity, and the representativeiness of the predict variables. ...
Effective water resources management in California relies substantially on real-time information of snow water equivalent (SWE) at basin and mountain range scales as mountain snowpacks provide the primary water supply for the State. However, SWE estimation based solely on remote sensing, modeling, or ground observations alone does not meet contemporary operational requirements. This study develops a statistically-based data-fusion framework to estimate SWE in real-time, which combines multi-source datasets including satellite-observed daily mean fractional snow-covered area (DMFSCA), snow pillow SWE measurements, physiographic data, and historical SWE patterns into a linear regression model (LRM). We test two LRMs: a baseline regression model (LRM-baseline) that uses physiographic data and historical SWE patterns as independent variables, and an FSCA-informed regression model (LRM-FSCA) that includes the DMFSCA from Moderate Resolution Imaging Spectroradiometer (MODIS) satellite imagery as an additional independent variable. The performance of the model is comprehensively evaluated and compared with two operational models – the Snow Data Assimilation System (SNODAS) and the National Water Model (NWM-SWE). By incorporating the satellite-observed DMFSCA, LRM-FSCA outperforms LRM-baseline with increased median R² from 0.54 to 0.60, and reduced median PBIAS of basin average SWE from 2.6% to 2.2% in the snow pillow cross-validation. LRM-FSCA explains 87% of the variance in the snow course SWE measurements with 0.1% PBIAS, while LRM-baseline explains a lower 81% variance with 1.4% PBIAS, both of which show higher accuracy than SNODAS (73% and -2.4%, respectively) and NWM-SWE (75% and -15.9%, respectively). Additionally, LRM-FSCA explains 85% of the median variance in the Airborne Snow Observatory SWE with -9.2% PBIAS, which is comparable to the LRM-baseline (86% and -11.3%, respectively) and substantially better than SNODAS (64% and 28.2%, respectively) and NWM-SWE (33% and -30.1% respectively). This study shows a substantial model improvement by constraining the geographical and seasonal variation on snow-cover via satellite observation and highlights the values of using multi-source observations in real-time SWE estimation. The developed SWE estimation framework has crucial implications for effective water supply forecasting and management in California, where climate extremes (e.g., droughts and floods) require particularly skillful monitoring practices.
... Although the alpine and subalpine areas evaluated are representative of mountainous terrain in the region and snowpacks in this area are representative of the continental snow regime (Trujillo and Molotch, 2014), further analysis of subgrid snow variability across a greater geographic area and across other regions with differing snow regimes could improve the applicability of a CVds parameterization for snow distributions in mountains areas in general. Additionally, spatial patterns of snow variability have been shown to be temporally consistent from year-to-year (Erickson et al., 2005;Deems et al., 2008;Sturm and Wagner, 2010], but future studies with multiple years of lidar collection could help understand the inter-annual variability of CVds and the consistency of its driving variables (Fassnacht et al., 2012]. Of particular interest would be the temporal consistency of the relation between CVds and ds. ...
Given the substantial variability of snow in complex mountainous terrain, a considerable challenge of coarse scale modeling applications is accurately representing the subgrid variability of snowpack properties. The snow depth coefficient of variation (CVds) is a useful metric for characterizing subgrid snow distributions but has not been well defined by a parameterization for mountainous environments. This study utilizes lidar-derived snow depth datasets spanning alpine to sub-alpine mountainous terrain in Colorado, USA to evaluate the variability of subgrid snow distributions within a grid size comparable to a 1000 m resolution common for hydrologic and land surface models. The subgrid CVds exhibited a wide range of variability across the 321 km2 study area (0.15 to 2.74) and was significantly greater in alpine areas compared to subalpine areas. Mean snow depth was the dominant driver of CVds variability in both alpine and subalpine areas, as CVds decreased nonlinearly with increasing snow depths. This negative correlation is attributed to the static size of roughness elements (topography and canopy) that strongly influence seasonal snow variability. Subgrid CVds was also strongly related to topography and forest variables; important drivers of CVds included the subgrid variability of terrain exposure to wind in alpine areas and the mean and variability of forest metrics in subalpine areas. Two statistical models were developed (alpine and subalpine) for predicting subgrid CVds that show reasonable performance statistics. The methodology presented here can be used for characterizing the variability of CVds in snow-dominated mountainous regions, and highlights the utility of using lidar-derived snow datasets for improving model representations of snow processes.
... As the elevation increases, the effect of human activities on vegetation weakens, and NDVI increases [48][49][50]. As the elevational gradient increases, the rate of NDVI growth decreases, which may stem from changes in natural environmental factors such as temperature, precipitation, and light [51][52][53]. Higher elevation corresponds to lower soil temperature, higher humidity, and greater vulnerability to leaching, which leads to soil acidification and inhibits the growth of vegetation [54]. ...
The temporal and spatial characteristics of vegetation in the middle reaches of the Yangtze River (MRYR) were analyzed from 1999 to 2015 by trend analysis, co-integration analysis, partial correlation analysis, and spatial analysis using MODIS-NDVI time series remote sensing data. The average NDVI of the MRYR increased from 0.72 to 0.80, and nearly two-thirds of the vegetation showed a significant trend of improvement. At the inter-annual scale, the relationship between NDVI and meteorological factors was not significant in most areas. At the inter-monthly scale, NDVI was almost significantly correlated with precipitation, relative humidity, and sunshine hours, and the effect of precipitation and sunshine hours on NDVI showed a pronounced lag. When the altitude was less than 2500 m, NDVI increased with elevation. NDVI increased gradually as the slope increased and decreased gradually as the slope aspect changed from north to south. NDVI decreased as the population density and per capita GDP increased and was significantly positively correlated with afforestation policy. These findings provide new insights into the effects of climate change and human activities on vegetation growth.
... Topographic parameters can be used as proxies for the meteorological drivers, such as precipitation or wind for sublimation and redistribution or solar radiation (and temperature) for snowmelt. In addition, vegetation, and in particular the presence and density of a canopy, affects local meteorological conditions [40]. Several works aim at understanding the relationship between snowpack distribution and properties, and topographic variables. ...
... With the purpose of producing SWE maps, Erxleben et al. [41] considered elevation, slope, aspect, and forest coverage. Since elevation and SWE are known to be highly correlated [4], Fassnacht et al. [40] examined the relation between SWE and other topographic parameters, including location, canopy density, slope and aspect. ...
This paper presents a new concept to derive the snow water equivalent (SWE) based on the joint use of snow model (AMUNDSEN) simulation, ground data, and auxiliary products derived from remote sensing. The main objective is to characterize the spatial-temporal distribution of the model-derived SWE deviation with respect to the real SWE values derived from ground measurements. This deviation is due to the intrinsic uncertainty of any theoretical model, related to the approximations in the analytical formulation. The method, based on the k-NN algorithm, computes the deviation for some labeled samples, i.e., samples for which ground measurements are available, in order to characterize and model the deviations associated to unlabeled samples (no ground measurements available), by assuming that the deviations of samples vary depending on the location within the feature space. Obtained results indicate an improved performance with respect to AMUNDSEN model, by decreasing the RMSE and the MAE with ground data, on average, from 154 to 75 mm and from 99 to 45 mm, respectively. Furthermore, the slope of regression line between estimated SWE and ground reference samples reaches 0.9 from 0.6 of AMUNDSEN simulations, by reducing the data spread and the number of outliers.
... Given the non-linear height dependence of SWE, observational data on high elevation levels are therefore subject to large uncertainties. This is a common issue on high elevation levels [30][31][32]. Also, many stations-typically those in Swiss winter tourism destinations-are shut down during the summer period. ...
The recent development of high-resolution climate models offers a promising approach in improving the simulation of precipitation, clouds and temperature. However, higher grid spacing is also a promising feature to improve the simulation of snow cover. In particular, it provides a refined representation of topography and allows for an explicit simulation of convective precipitation processes. In this study we analyze the snow cover in a set of decade-long high-resolution climate simulation with horizontal grid spacing of 2.2 km over the greater Alpine region. Results are compared against observations and lower resolution models (12 and 50 km), which use parameterized convection. The simulations are integrated using the COSMO (Consortium for Small-Scale Modeling) model. The evaluation of snow water equivalent (SWE) in the simulation of present-day climate, driven by the ERA-Interim reanalysis, against an observational dataset, reveals that the high-resolution simulation clearly outperforms simulations with grid spacing of 12 and 50 km. The latter simulations underestimate the cumulative amount of SWE over Switzerland over the whole annual cycle by 33% (12 km simulation) and 56% (50 km simulation) while the high-resolution simulation shows a spatially and temporally averaged difference of less than 1%. Scenario simulations driven by GCM MPI-ESM-LR (2081-2090 RCP8.5 vs. 1991-2000) reveal a strong decrease of SWE over the Alps, consistent with previous studies. Previous studies had found that the relative decrease becomes gradually smaller with elevation, but this finding was limited to low and intermediate altitudes (as a 12 km simulation resolves the topography up to 2500 m). In the current study we find that the height gradient reverses sign, and relative reductions in snow cover increases above 3000 m asl, where important parts of the cryosphere are present. In addition, the simulations project a transition from permanent to seasonal snow cover at high altitudes, with potentially important impacts to Alpine permafrost. This transition and the more pronounced decline of SWE emphasize the value of the higher grid spacing. Overall, we show that high-resolution climate models offer a promising approach in improving the simulation of snow cover in Alpine terrain.
... The statistical analyses identified the terrain and canopy variables that describe the distribution and variability of snow depth about the two SNOTEL stations. These terrain variables included eleva- (Dingman et al., 1988;Fassnacht et al., 2003), slope as the sine of slope (Sexstone and Fassnacht, 2014), northness (Fassnacht et al., 2012), eastness (Wallace and Gass, 2008), cumulative monthly clear sky solar radiation (Dozier and Frew, 1990), and maximum upwind slope Fig. 3 Mean absolute difference from the pixel mean snow depth for all samples per pixel as a function of the number of points per pixel for the three sampling dates. The number in the (a) parentheses is the number of samples per pixel, while (b) is for only the 11 points in a row (as per Fig. 1(c)-ii). ...
Snow depth is the easiest snowpack property to measure in the field and is used to estimate the distribution of snow for quantifying snow storage. Often the mean of three snow depth measurements is used to represent snow depth at a location. This location is used as a proxy for an area, typically a digital elevation model (DEM) or remotely sensed pixel. Here, 11, 17, or 21 snow depth measurements were used to represent the mean snow depth of a 30-m DEM pixel. Using the center snow depth measurement for each sampling set was not adequate to represent the pixel mean, and while the use of three snow depth measurements improved the estimate of mean, there is still large error for some pixels. These measurements were then used to determine the variability of snow depth across a pixel. Estimating variability from few points rather than all in a measurement was not sufficient. The sampling size was increased from one to the total per pixel (11, 17, or 21) to determine how many point samples were necessary to approximate the mean snow depth per pixel within five percent. Binary regression trees were constructed to determine which terrain and canopy variables dictated the spatial distribution of the snow depth, the standard deviation of snow depth, and the sample size to within 5% of the mean per pixel. One location was measured in two years just prior to peak accumulation, and it is shown that there was little to no inter-annual consistency in the mean or standard deviation.
... A mix of static physiographic (Fassnacht et al., 2012) and dy- namic variables were used as predictors (Table 2). All vari- ables were computed at or resampled to 3.125 km resolution using Gaussian pyramid reduction or expansion (Burt and Adelson, 1983) for the initial steps and bilinear interpola- tion for the final step. ...
... All other predictors show importance < 0.08. The third most important predictor is elevation, shown to be an im- portant predictor in previous studies (Fassnacht et al., 2003;Fassnacht et al., 2012;Schneider and Molotch, 2016). The fourth most important variable is longitude, followed by TB 18V -TB 36V , the difference between microwave brightness temperatures at 18 and 36 GHz, showing that the passive mi- crowave SWE retrievals have little predictive power. ...
In the mountains, snowmelt often
provides most of the runoff. Operational estimates use imagery from optical
and passive microwave sensors, but each has its limitations. An accurate
approach, which we validate in Afghanistan and the Sierra Nevada USA,
reconstructs spatially distributed snow water equivalent (SWE) by calculating
snowmelt backward from a remotely sensed date of disappearance. However,
reconstructed SWE estimates are available only retrospectively; they do not
provide a forecast. To estimate SWE throughout the snowmelt season, we
consider physiographic and remotely sensed information as predictors and
reconstructed SWE as the target. The period of analysis matches the AMSR-E
radiometer's lifetime from 2003 to 2011, for the months of April through
June. The spatial resolution of the predictions is 3.125 km, to match the
resolution of a microwave brightness temperature product. Two machine
learning techniques – bagged regression trees and feed-forward neural
networks – produced similar mean results, with 0–14 % bias and
46–48 mm RMSE on average. Nash–Sutcliffe efficiencies averaged 0.68 for
all years. Daily SWE climatology and fractional snow-covered area are the
most important predictors. We conclude that these methods can accurately
estimate SWE during the snow season in remote mountains, and thereby provide
an independent estimate to forecast runoff and validate other methods to
assess the snow resource.
... [40,41]. Observations were disproportionately sparse at the higher elevations of the RMNP (Figure 1b), which is common for such stations [42,43]. The RMNP extends from 2100 m up to a maximum elevation of 4346 m (Longs Peak), but approximately 1/3 of the RMNP's hypsometry is within the alpine (>3505 m) and approximately 50% is within the sub-alpine or zone of maximum seasonal snow cover (2895-3505 m; Figure 1). ...
We present a detailed study of the snowpack trends in the Rocky Mountain National Park (RMNP) using snow telemetry and snow course data at a monthly resolution. We examine the past 35 years (1981 to 2016) to explore monthly patterns over 36 locations and used some additional data to help interpret the changes. The analysis is at a finer spatial and temporal scale than previous studies that focused more on aggregate- or regional-scale changes. The trends in the first of the month’s snow water equivalent (SWE) varied more than the change in the monthly SWE, monthly precipitation or mean temperature. There was greater variability in SWE trends on the west side of the study area, and on average the declines in the west were greater. At higher elevations, there was more of a decline in the SWE. Changes in the climate were much less in winter than in summer. Per decade, the average decline in the winter precipitation was 4 mm and temperatures warmed by 0.29 °C, while the summer precipitation declined by 9 mm and temperatures rose by 0.66 °C. In general, November and March became warmer and drier, yielding a decline of the SWE on December 1st and April 1st, while December through February and May became wetter. February and May became cooler.
