Chapter 4: Changes in Snowpack & Snow Residence Time

Dataset Overview 

This dataset includes modeled historical (1975-2005) and future (2071-2090, RCP 8.5) projections of April 1st snow-water equivalent (SWE) and snow-residence time (SRT) (figure 4.1; see chapter glossary for definitions of terms). The model that produced these datasets uses average winter (November through March) temperature and total winter precipitation. The model was validated using data from 497 Snow Telemetry (SNOTEL) sites in the western United States. Changes in SWE and SRT are presented both as absolute changes (future values minus historical values) and as percent changes. 

Data Access: https://www.fs.fed.us/rm/boise/AWAE/projects/NFS-regional-climate-change-maps/categories/us-raster-layers.html

Figure 4.1

 

Conservation Applications

Potential conservation applications of this dataset could include the following:

  • This dataset can be used to envision and plan for future losses in winter snowpack. Winter snowpack reductions can ripple through the hydrologic system of watersheds throughout the year (e.g., contributing to reduced streamflow in summer). As a result, this dataset is potentially relevant to several conservation questions related to human and ecological water use. In areas where fish habitat quality and human water withdrawals are strongly dependent on streamflow that derives from snowmelt, these projections of snowpack dynamics could be used to begin planning for the implications of snowpack reduction. These changes could have implications for fire management, recreation, and agriculture, in addition to ecological effects on snow-dependent plants and animals.

Applicable scales for detailed spatial assessments:

  • For conservation applications requiring detailed assessment of spatial variation of the dataset within a geographic boundary (such as a protected area), the following geographic scales may be most appropriate (see appendix 3 for more information).
     
  • At the scale of: a level-3 ecoregion (e.g., the North Cascades), a single state (Washington, Oregon, or Idaho), a region (multiple states in the Pacific Northwest or in western North America), the continental United States.

Applicable scales for assessing general patterns:

  • Due to spatial resolution, the dataset may not show detailed spatial variation at the following geographic scales, however, the dataset may be useful to assess general patterns or for comparison to other locations (see appendix 3 for more information).
     
  • At the scale of: a small (< 1 km2) nature preserve, a state park or state wildlife area, a local watershed (12-digit hydrologic unit code [HUC-12]), a Bureau of Land Management (BLM) district, a river watershed (8-digit hydrologic unit code [HUC-8]), an individual county, a national forest.

Use of the dataset in conservation applications may be limited by the following considerations:

  • This dataset portrays a future of snowpack reductions averaged over several years but does not represent variation in snowpack dynamics from one year to another. It is an independent snow model, so predicting changes in streamflow resulting from changes in snowpack dynamics would require additional hydrologic modeling. Furthermore, because of its spatial resolution and the nature of the input data used to produce the data layers, this dataset cannot be used to evaluate fine-scale differences in snowpack dynamics, such as those produced by topographic variation in rough terrain.

Past or current conservation applications:

  • The dataset has been used in vulnerability studies and forest planning in several national forests across the Northwest, details of which may be obtained by contacting the corresponding author. 

Dataset citation:

Luce, C., V. Lopez-Burgos, and Z. Holden. 2014. Sensitivity of snowpack storage to precipitation and temperature using spatial and temporal analog models. Water Resources Research 50:2108–2123.

Lute, A., and C. Luce. 2017. Are model transferability and complexity antithetical? Insights from validation of a variable-complexity empirical snow model in space and time. Water Resources Research 53:8825–8850.

Dataset documentation links:

https://doi.org/10.1002/2013WR014844 (open access)

https://doi.org/10.1002/2017WR020752 (subscription or fee required)

Data access:

The dataset can be downloaded from: https://www.fs.fed.us/rm/boise/AWAE/projects/NFS-regional-climate-change-maps/categories/us-raster-layers.html

Forest-specific maps are also available: https://www.fs.fed.us/rm/boise/AWAE/projects/national-forest-climate-change-maps.html

These data can be viewed online at the U.S. Department of Agriculture (USDA) Forest Service Geospatial Data Discovery page, including historical, future, absolute change, and percent change grids (rasters) for SRT and SWE:

https://enterprisecontent-usfs.opendata.arcgis.com/datasets?q=snow%20residence%20time

https://enterprisecontent-usfs.opendata.arcgis.com/datasets?q=snow%20water%20equivalent

These are also accessible as web services: https://apps.fs.usda.gov/fsgisx01/rest/services/RDW_Climate

A story map was created by the U.S. Forest Service Office of Sustainability and Climate to present and compare these datasets: https://usfs.maps.arcgis.com/apps/MapSeries/index.html?appid=4d6e58342f5a451dbe9e9c946bf76f85

Metadata access:

Formal metadata can be downloaded from: https://www.fs.fed.us/rm/boise/AWAE/projects/NFS-regional-climate-change-maps/downloads/NationalForestClimateChangeMapsMetadata.pdf

Dataset corresponding author:
Charles Luce
USDA Forest Service, Rocky Mountain Research Station
[email protected]

Data type category (as defined in the Introduction to this guidebook): Hydrology, climate, (winter) animal habitat

Species or ecosystems represented: This dataset does not represent any individual species or ecosystems.

Units of mapped values:

Historical April 1st SWE: millimeters (mm)
Future April 1st SWE: mm
Absolute change in April 1st SWE: mm
Percent change in April 1st SWE: percent
Historical SRT: days
Future SRT: days
Absolute change in SRT: days
Percent change in SRT: percent

Range of mapped values:

Historical April 1st SWE: 0 to 3,731.08
Future April 1st SWE: 0 to 2,579.12
Absolute change in April 1st SWE: -2,214.3 to 164.847
Percent change in April 1st SWE: -100 to 11.7575
Historical SRT: 0 to 193.961
Future SRT: 0 to 165.841
Absolute change in SRT: -115.496 to 0
Percent change in SRT: -100 to -3.20848

Spatial data type: a raster dataset (grid)

Data file format(s): GeoTiff (.tif)

Spatial resolution: 1/24 degree

Geographic coordinate system: World Geodetic System (WGS) 1984

Spatial extent: Conterminous United States

Dataset truncation: The dataset is truncated along the United States borders with Canada and Mexico.

Time period represented: Historical (1975-2005); future (later than 2020)

Future time period(s) represented: End-of-century (2071-2090)

Baseline time period (against which future conditions were compared): 1975-2005.

Methods overview:

First, a model was developed using cumulative winter (November through March) precipitation and average winter temperature to predict April 1st SWE and SRT (Luce et al. 2014). The model was calibrated and verified using data from approximately 500 SNOTEL sites. As documented in Lute and Luce (2017), models of varying complexity (i.e., models with varying numbers of parameters to be adjusted by calibration) were compared and it was found that the simpler models (with fewer parameters) generally outperformed the more complex models. Therefore, a relatively simple model was chosen to create the spatial layers of future SWE and SRT. Using that model, gridded projections of future winter temperature and precipitation from the MACAv2-Metdata dataset (Abatzoglou and Brown 2012) were used to create gridded projections of April 1st SWE and SRT. For more information, please consult the dataset citation listed in section 2 of this chapter.

This dataset employed the following specific models: locfit package in the R software environment (Loader, 2013)

Major input data sources for this dataset included: Historical climate observations or models, future climate projections, hydrologic data (snowpack variables)

This dataset used the following general circulation models (GCMs): BCC-CSM1-1, BCC-CSM1-1-M, BNU-ESM, CanESM2, CCSM4, CNRM-CM5, CSIRO-Mk3.6.0, GFDL-ESM2M, GFDL-ESM2G, HadGEM2-ES, HadGEM2-CC, INM-CM4, IPSL-CM5A-LR, IPSL-CM5A-MR, IPSL-CM5B-LR, MIROC5, MIROC-ESM, MIROC-ESM-CHEM, MRI-CGCM3, NorESM1-M

An ensemble across GCMs is provided; individual files for individual GCMs are not provided. More information about climate models is available in appendix 1. Detailed information about climate models, including model evaluation and comparison among models, is available from Randall et al. (2007) and Rupp et al. (2013).

This dataset used the following greenhouse-gas scenarios: RCP 8.5

More information about greenhouse-gas scenarios is available in appendix 2 and from Knutti and Sedláček (2013).

Creation of this dataset involved the following methods to change the spatial resolution of climate models (e.g. to downscale or resample climate models):

The Multivariate Adaptive Constructed Analogs (MACA) downscaling method was used (Abatzoglou and Brown, 2012).

The mapped values of the dataset may be interpreted as follows:

SWE and SRT are measured in mm and days, respectively, for both the historical and future time periods. Higher SWE values indicate more snow present on April 1st; areas with higher SRT values are projected to have snow retained for a larger number of days each winter. For the absolute change rasters, large negative values in April 1st SWE indicate the largest April 1st SWE decreases. A few areas show positive values for this layer, representing projected increases in April 1st SWE. Similarly, large negative values of absolute change in SRT indicate the largest decreases in the amount of time each winter that snow is expected to be present. For both April 1st SWE and SRT, the lowest possible percent value is -100, which indicates complete loss of snowpack in the future.

Representations of key concepts in climate-change ecology:

The data layers in this dataset primarily represent one important component of climate-change exposure, namely the degree to which areas are expected to experience reduced winter snowpack. In some areas of the Pacific Northwest, transitions from rain to snow and earlier seasonal timing of snowmelt are major concerns related to climate-change effects on ecosystems and watershed hydrology. Snow conditions are ecologically important to a number of plant and animal species in the Pacific Northwest, especially those adapted to mountain habitats (e.g., alpine and subalpine species). For animal species, the seasonal timing, depth, and spatial extent of snowpack can affect access to food resources, ability to hide or escape from predators, and ease of movement across the landscape. For plant species, these snowpack characteristics can affect growth rates, seasonal availability of soil moisture, and seedling establishment. In addition, in watersheds where streams have large contributions from snowmelt, changes in snowpack patterns can have important consequences for watershed hydrology, including changes in timing of peak streamflow or reductions in late-season streamflow.

Because of these concerns, anticipated reductions in snowpack may need to be considered along with information on species and ecosystem-level sensitivity to snowpack changes and information on species’ capacity to adapt to changing snowpack conditions. Projected snowpack declines across much of the Pacific Northwest illustrate the importance of changes in watershed hydrology as a critical linkage between climate change and effects on species and ecosystems.

This dataset involves the following assumptions, simplifications, and caveats:

In the study that produced this dataset, model validations were better for April 1st SWE than for SRT, and the dataset authors suggest this may be because there is a stronger physical basis for modeling SWE than SRT (Lute and Luce, 2017). These models used only climate variables to predict changes in snowpack and the input climate variables were at 4-km resolution. As a result, finer-scale influences on snowpack dynamics, such as those based on microclimates produced by topography, are not represented in these data layers. Additionally, this dataset does not represent sensitivity or adaptive capacity, in terms of the ways in which species or ecosystems are expected to respond to projected snowpack reductions.

Quantification of uncertainty:

Several uncertainty metrics of the underlying snow models are provided in Lute and Luce (2017). Nash-Sutcliffe efficiency (NSE) values exceeded 0.71 for SWE and 0.64 for SRT (NSE = 1 indicates a perfect match between a model and observations, whereas NSE = 0 indicates model predictive ability as accurate as the mean of the observed data). Using all data, the site-wise cross validation was 85% for SWE and 81% for SRT. Although uncertainty of the mapped values (i.e., changes in SRT and April 1st SWE) are not provided with the maps, information necessary for calculating them is available; see Lute and Luce (2017).

Field verification:

Modeled snow variables from the historical (baseline) period were field verified (Lute and Luce 2017). Field verification of the projected mapped values in this dataset was not possible because the dataset represents a future condition. Two-fold validations in Lute and Luce (2017) were designed to assess the capability of the model to extrapolate to conditions other than those of calibration.

Prior to dataset publication, peer review was conducted by external review (at least two anonymous reviewers, each from a different institution). The underlying models were published in two peer-reviewed publications.

Abatzoglou, J. T., and T. J. Brown. 2012. A comparison of statistical downscaling methods suited for wildfire applications. International Journal of Climatology 32:772–780.

Knutti, R., and J. Sedláček. 2013. Robustness and uncertainties in the new CMIP5 climate model projections. Nature Climate Change 3:369–373.

Loader, C. 2013. Locfit: Local regression, likelihood and density estimation, in R package version 1.5–9.1. http://cran.rproject.org/package5locfit.

Luce, C., V. Lopez-Burgos, and Z. Holden. 2014. Sensitivity of snowpack storage to precipitation and temperature using spatial and temporal analog models. Water Resources Research 50:2108–2123.

Lute, A., and C. Luce. 2017. Are model transferability and complexity antithetical? Insights from validation of a variable-complexity empirical snow model in space and time. Water Resources Research 53:8825–8850.

Randall, D., R. Wood, S. Bony, R. Colman, T. Fichefet, J. Fyfe, V. Kattsov, and et al. 2007. Cilmate models and their evaluation. in S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K. Averyt, M. Tignor, and H. Miller, editors. Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, New York, NY.

Rupp, D. E., J. T. Abatzoglou, K. C. Hegewisch, and P. W. Mote. 2013. Evaluation of CMIP5 20th century climate simulations for the Pacific Northwest USA. Journal of Geophysical Research 118: 884–907.