Skip to main content
  1. ScienceBase Catalog
  2. National and Regional Climate Adaptation Science Centers
  3. Northwest CASC
  4. FY 2019 Projects
  5. Estimating the Spatial and Temporal Extent of Snowpack Properties in Complex Terrain: Leveraging Novel Data to Adapt Wildlife and Habitat Management Practices to Climate Change

Estimating the Spatial and Temporal Extent of Snowpack Properties in Complex Terrain: Leveraging Novel Data to Adapt Wildlife and Habitat Management Practices to Climate Change

Principal Investigator
Timothy Link

Dates

Release Date
2019
Start Date
2019-09-01
End Date
2024-09-30

Summary

Snow conditions are changing dramatically in the mountains of the interior Pacific Northwest, including eastern Washington, northern Idaho, and western Montana. These changes can both benefit and hinder a variety of wildlife species. The timing and extent of seasonal snowpacks, in addition to snow depth, density, and hardness, can impact the ability of wildlife to access forage, their ability to move across the landscape, and their vulnerability to predators, to name a few. In order to respond effectively to changes in snow conditions, wildlife managers need tools to identify areas and promote conditions that maintain late spring and early summer snowpack for some sensitive species. Managers also require an index of winter severity [...]

Child Items (3)

Contacts

Principal Investigator :
Timothy Link, David Ausband
Cooperator/Partner :
Adrienne Marshall, Kaitlyn Strickfaden, Leona K Svancara
Funding Agency :
Northwest CASC
CMS Group :
Climate Adaptation Science Centers (CASC) Program

Attached Files

Click on title to download individual files attached to this item.

Tree Camera - Dave Ausband.jpg
“Checking a camera trap at a field site; Photo Credit: Dave Ausband (USGS)” 		thumbnail 36.32 KB image/jpeg

Purpose

Snow conditions are changing dramatically in the mountains of the interior Pacific Northwest, including eastern Washington, northern Idaho, and western Montana. These changes can both benefit and hinder a variety of wildlife species. To respond effectively to climate changes and variations, wildlife managers need tools to identify areas and promote conditions that maintain late spring/early summer snowpack. Managers also require an index of winter severity based on temperature, snow depth, and snow hardness at relevant spatial and temporal scales to adapt management strategies for seasonal conditions. We propose to advance the understanding of how snow conditions vary and how such variation affects both species of greatest conservation need (e.g., wolverine, hoary marmot, western bumble bee, mountain goat) and species of economic and recreational importance (e.g., elk, moose) in forests spanning the rain-snow transition zone in the interior Pacific Northwest. We will achieve these outcomes by creating new tools that managers can use to identify areas of late season snow and estimate winter severity from remote sensing data and automated cameras. We will then use this novel data to predict habitat use by selected species and survival of ungulates in the region. Providing natural resource managers with tools to identify locations of snow retention for sensitive and listed species is critical for identifying habitats to conserve and/or modify (e.g. small canopy gap creation) for species recovery. Lastly, a winter severity model will provide wildlife managers with a much-needed tool for assessing climate change effects on ungulates and adjusting management strategies (e.g., harvest reduction, securing more winter range habitat) accordingly.

Project Extension

projectStatusIn Progress

Checking a camera trap at a field site; Photo Credit: Dave Ausband (USGS)
Checking a camera trap at a field site; Photo Credit: Dave Ausband (USGS)

Map

Map Preview

Spatial Services

ScienceBase WMS

Communities

  • National and Regional Climate Adaptation Science Centers
  • Northwest CASC

Tags

Provenance

[:]-[:]

Additional Information

Data Management Plan Extension

customSoftware
descriptionThis R package can extract edges of a snow stake in "reference images" and superimpose those edges on to new images, giving a way to measure snow depth without needing to leave equipment in the camera viewshed.
repositoryThe R package is hosted on GitHub.
webToolMaintenanceAndSupportThe R package is hosted on GitHub. Issues posted on the GitHub page will be addressed as they are brought up by users of the R package.
languagesProgram R
restrictionsNo restrictions.
backupAndStorageThe R package is hosted on GitHub, which tracks changes to the software.
nameedger: An R Package Facilitating Distance and Snow Depth Measurements at Remote Camera Stations
dataProduct
metadataFGDC
exclusiveUseAll of these data will be available through published results and in the thesis of the Masters student that is on the project. All UI thesis are available electronically through the UI library.
descriptionMap of predicted relative snow disappearance dates across Moscow Mountain in Latah County, ID.
repositoryNone planned at this time.
format.tif, .jpg
restrictionsNo restrictions.
backupAndStorageThe shapefile and image file of predicted snow disappearance dates are backed up on an external hard drive.
dataManagementResourcesData will be housed in an OwnCloud directory structure maintained by the Northwest Knowledge Network. $1000 (.33%) of the budget are allocated to cover NKN fees in addition to M.S. student time required for data management tasks and technician time in maintaining cameras and processing images.
volumeEstimate5 MB
nameSnow Refugia Map
history2022-10-17 18:53:55 MDT: phase Submitted DMP
newInput
metadataFGDC
exclusiveUseNo limitations.
descriptionHemispherical photographs taken at remote camera stations on Moscow Mountain in Latah County, ID. Hemispherical photographs were taken to quantify incoming shortwave radiation at the camera sites. Photographs were analyzed using the attributes listed in the .csv file in Hemisfer software.
repositoryNone planned at this time.
qualityChecksHemispherical photographs were only taken at dawn and dusk or on overcast days to ensure good lighting conditions. The hemispherical camera was carefully levelled for each hemispherical photograph. Settings in Hemisfer were tailored to each image to ensure proper thresholding. Images were edited when necessary to remove an object which was interfering with thresholding.
protocolsWe took hemispherical photographs (hemiphotos) at each camera station to estimate incoming shortwave radiation. We used a Canon EOS 70D SLR camera with a Sigma 8mm circular fisheye lens. We took hemiphotos on days with little to no wind or precipitation and early in the morning (5AM – 7AM), late in the evening (7PM – 9PM), or on overcast days to maximize the contrast between sky and vegetation. We took hemiphotos in June - July 2021, but nine camera sites were re-photographed in October 2021 due to poor quality of initial photographs. Overstory deciduous vegetation is rare at the study site, so summer changes in canopy cover are negligible. At each camera site, we placed the DSLR camera with attached lens on a tripod and levelled it to point directly up into the canopy. One set of hemiphotos was taken at the tree to which the camera was mounted, and a second was taken 5 m into the camera viewshed. We took photographs at multiple exposures using the auto-exposure bracketing settings of the camera. We analyzed hemiphotos using Hemisfer software. We selected the hemiphoto with the best exposure for each location at a camera site. We manually selected threshold values and adjusted color-weighting when needed to distinguish sky pixels from vegetation pixels in each image. Hemisfer then classified sky and vegetation pixels in the photograph using the threshold, overlaid a solar path onto photographs based on input georeferencing information and photograph orientation. We set Hemisfer to calculate hourly direct and diffuse shortwave radiation throughout time assuming 50% cloud cover. Hemisfer also output a count of sky and vegetation pixels in the image which we used to calculate percent vegetation cover.
format.jpg, .csv
restrictionsNo restrictions.
backupAndStorageHemispherical photographs and parameter files are backed up on an external hard drive.
dataManagementResourcesData will be housed in an OwnCloud directory structure maintained by the Northwest Knowledge Network. $1000 (.33%) of the budget are allocated to cover NKN fees in addition to M.S. student time required for data management tasks. Hemispherical photographs were mostly taken in summer 2021, though a few were re-photographed in fall 2021.
volumeEstimate1.8 GB
dataProcessingHemiphotos were processed using Hemisfer software.
nameHemispherical Photographs from Moscow Mountain (Latah County, ID)
metadataFGDC
exclusiveUseNo limitations.
descriptionSnow hardness data collected with a ram penetrometer near remote camera stations on Moscow Mountain in Latah County, ID.
repositoryNone planned at this time.
qualityChecksA sample was retaken if a stick or rock was encountered in the snowpack or if the ramsonde slipped horizontally for any other reason. An entry was marked as "DROP" if the resistance measurement was unusually high, particularly if that entry occurred at the base of the snowpack where it was likely a recording of the hardness of the underlying soil.
protocolsWe took snow density and hardness measurements at camera sites from December 2020 - April 2021. We took measurements every few weeks as logistics allowed. We took density and hardness samples near the camera site in snow visually similar to the snow in the camera viewshed to prevent snow conditions from being disturbed beyond normal camera deployment. We measured snow hardness using a ram penetrometer or “ramsonde” (Snowmetrics; Fort Collins, CO). A ramsonde is composed of a hammer, anvil, and rod. The hammer is dropped onto the anvil from a known height, and the depth to which the ramsonde penetrates the snow is recorded using gradations on the rod. Drops are performed until the tip of the ramsonde reaches the ground. Based on the weight of the ramsonde, the number of drops, and the height of the drops, a ram resistance value can be calculated, which serves as a proxy for snow hardness. We calculated ram resistance for the individual snow layers in the three hardness samples collected on each sampling occasion.
format.csv
restrictionsNo restrictions.
backupAndStorageSnow hardness data are backed up on an external hard drive.
dataManagementResourcesData will be housed in an OwnCloud directory structure maintained by the Northwest Knowledge Network. $1000 (.33%) of the budget are allocated to cover NKN fees in addition to M.S. student time required for data management tasks and field technician time for conducting fieldwork. Snow hardness data were collected between December 2020 - April 2021.
volumeEstimate729 KB
dataProcessingData were entered into Microsoft Excel.
nameSnow Hardness on Moscow Mountain (Latah County, ID)
metadataFGDC
exclusiveUseNo limitations.
descriptionSnow density data collected with a federal or prairie snow sampler near remote camera stations on Moscow Mountain in Latah County, ID.
repositoryNone planned at this time.
qualityChecksWe retained the core if the depth of snow in the sampler was at least 90% of the actual snow depth and the base of the snowpack had been reached as evidenced by litter or a soil plug at the base of the core. Data were checked by the graduate student on the project to detect unusual or impossible values.
protocolsWe took snow density and hardness measurements at camera sites from December 2020 - April 2021. We took measurements every few weeks as logistics allowed. We took density and hardness samples near the camera site in snow visually similar to the snow in the camera viewshed to prevent snow conditions from being disturbed beyond normal camera deployment. We took snow density samples using a homemade prairie sampler in snow depths < 100 cm and using a federal snow sampler in snow depths > 100 cm. The sampler was inserted into the snow to remove a snow core. We retained the core if the depth of snow in the sampler was at least 90% of the actual snow depth and the base of the snowpack had been reached as evidenced by litter or a soil plug at the base of the core. After we removed the soil plug, we weighed the core to determine its snow-water equivalent (SWE). We converted the SWE measured with the samplers into a density measurement by dividing the SWE by the snow depth. If a snow core of adequate quality could not be obtained after several minutes of effort, we did not measure snow density on that sampling occasion.
format.csv
restrictionsNo restrictions.
backupAndStorageSnow density data are backed up on an external hard drive.
dataManagementResourcesData will be housed in an OwnCloud directory structure maintained by the Northwest Knowledge Network. $1000 (.33%) of the budget are allocated to cover NKN fees in addition to M.S. student time required for data management tasks and field technician time for conducting fieldwork. Snow density data were collected Between December 2020 - April 2021.
volumeEstimate113 KB
dataProcessingData were entered into Microsoft Excel.
nameSnow Density on Moscow Mountain (Latah County, ID)
metadataFGDC
exclusiveUseNo limitations.
descriptionAttributes of remote camera stations on Moscow Mountain in Latah County, ID including georeferencing information and camera deployment information.
repositoryNone planned at this time.
qualityChecksData sheets were filled out while deploying cameras. Slope and aspect were determined using a 1-m digital elevation model to ensure precision in values. Cameras were checked throughout deployment for proper function. Cameras were locked onto trees with cable locks to prevent theft. LogTags were hung from the same heights as cameras to ensure that temperature differences were not the result of elevational gradients in temperature. Hemispherical photographs were only taken at dawn and dusk or on overcast days to ensure good lighting conditions. The hemispherical camera was carefully levelled for each hemispherical photograph. Settings in Hemisfer were tailored to each image to ensure proper thresholding. Images were edited when necessary to remove an object which was interfering with thresholding.
protocolsWe used stratified sampling to select camera sites. Stratifications were based on elevation (800-1500 m above sea level, in 5 bins), aspect (N, S, E, or W), and canopy cover (<35% = sparse, 35-75% = moderate, or >75% = dense). We hiked along roads or trails and went off-trail when we arrived at an area at an elevation and aspect that we needed to sample. We searched for trees with a view of an area that had little obstructing vegetation. We then used a densiometer to measure the canopy cover in this area. If this area met criteria we needed for sampling, we deployed a camera station in this area. We deployed a camera and a LogTag at every site. We deployed a snow stake at a subset of sites close to major roads to minimize field effort. The camera was mounted to a tree 2-3 m from the ground, and the LogTag was hung on a branch at the same height in a shaded area. On deployment, the latitude, longitude, and elevation of the site were determined using a Garmin Oregon 500 GPS unit. We also recorded the aspect and canopy cover, the height and direction of the camera, and the prevailing understory cover. We took reference images for superimposing a "virtual" snow stake using the edger package in program R. We placed a snow stake at 5, 10, and 15 m into the viewshed of the camera and allowed the camera to take motion-triggered images. When possible, a technician stood behind the snow stake to maximize contrast. We removed the snow stakes when we finished deploying the camera. We also removed any vegetation which might obstruct the view of the camera. We retrieved cameras in April and May 2021. We retrieved a camera if all snow at the camera site had melted. However, due to logistical constraints, a few cameras had to be retrieved when a small amount of snow was still present at the sites. We took additional reference images during camera retrieval in case the camera's viewshed changed during its deployment.
format.csv
restrictionsNo restrictions.
backupAndStorageCamera station data have been backed up on an external hard drive.
dataManagementResourcesData will be housed in an OwnCloud directory structure maintained by the Northwest Knowledge Network. $1000 (.33%) of the budget are allocated to cover NKN fees in addition to M.S. student time required for data management tasks and field technician time for conducting fieldwork. Approximately 1.5 months were needed to deploy all cameras in fall 2020, then another approximately 1.5 months were needed to retrieve all cameras in spring 2021.
volumeEstimate25 KB
dataProcessingData were entered in Microsoft Excel. Elevation, slope, and aspect were found in ArcMap. Hemispherical photographs were processed in Hemisfer software.
nameAttributes of Remote Camera Stations on Moscow Mountain (Latah County, ID)
metadataFGDC
exclusiveUseAll of these data will be available through published results and in the thesis of the MS student on the project. All UI thesis are available electronically through the UI library.
descriptionHourly and motion-triggered images were collected from Fall 2020 through Spring 2021 at 5 elevation categories (800-925m, 925-1050m, 1050-1175m, 1775-1300m, and > 1300m), 4 aspects (N, S, E, and W), and 3 canopy densities (Sparse [0-35%], Moderate [35-75%], and Dense [75-100%]), in duplicate, plus 17 selected microclimates (137 locations total), on Moscow Mountain in Latah County, ID. Images from 27 other locations were part of a pilot experiment during January - May 2020. Data in the CSVs includes image metadata, camera site characteristics, temperature (degrees Celsius), precipitation events (T/F), snow presence (T/F), manual measurements of snow depth (cm), and wildlife detections.
repositoryNone planned at this time.
qualityChecksCamera sites were selected using strict criteria. Cameras were not deployed on trails and roads to prevent snow depths from being biased low. Cameras were tilted downward to prevent snow from accumulating on the lens which would prevent high-quality images from being taken. Cameras were checked regularly to ensure proper function. As images were processed, the MS student reviewed CSVs generated by technicians to find suspect values entered into the CSVs (text values in numeric fields, unreasonably large or small snow depth estimates, missing or unreasonable wildlife identifications and counts, etc.).
protocols"Camera sites were chosen by stratified non-random sampling. Field technicians were responsible for choosing sites that met the desired elevation, aspect, and canopy cover criteria. Cameras were never closer than 25m to other cameras, nor were they placed facing trails. Cameras were deployed so they had at least a 10-m unobstructed field of view (FOV). Branches and vegetation which could impede the FOV of the camera or cause false-positive triggers were removed. Cameras were deployed approximately 2m from the ground on trees and tilted slightly downward to prevent snow from accumulating on the lens. Cameras used included Reconyx (R) Hyperfire and Hyperfire II cameras. Cameras were programmed to take hourly photographs as well as motion-triggered photographs at high sensitivity. Cameras were programmed to take three images per trigger with a 1-second delay between images and no quiet period between triggers. Cameras were checked approximately monthly to ensure proper function. After collection of cameras, images were pre-processed to superimpose a ""virtual"" snow stake (VSS) onto images for snow depth estimations. The VSS method was developed by the MS student in Program R and allows the user to superimpose a snow stake onto images based on reference images taken during camera deployment or retrieval which contained a snow stake. Pre-processed images were then manually processed by technicians. Technicians measured snow depth using VSS's at 5m, 10m, and 15m from the camera, dependent on the camera viewshed. A subset of cameras (20) also had a permanent PVC snow stake installed in the camera viewshed. Technicians also reported snow presence, precipitation events (either happening [True] or not happening [False]), and wildlife detections. Wildlife detections were recorded such that every image in which a particular wildlife species appeared was recorded as ""Present."" Then, each individual was counted a single time in the proper demographic category (female/antlerless male, male, fawn, and unknown for ungulate species and adult or young-of-year for predator species). Other data recorded included camera operating state (normal or otherwise), human detections, and unique markings on wildlife."
format.csv
restrictionsOnly CSVs of data extracted from images will be made publicly available, but image data may be provided upon request.
backupAndStorageImage and CSV data have been backed up on three external hard drives.
dataManagementResourcesData will be housed in an OwnCloud directory structure maintained by the Northwest Knowledge Network. $1000 (.33%) of the budget are allocated to cover NKN fees in addition to M.S. student time required for data management tasks.
volumeEstimate193 MB
dataProcessingImages were manipulated using Program R. Images were processed by technicians using Timelapse2 software.
nameAutomated Image Database from Moscow Mountain (Latah County, ID)
metadataFGDC
exclusiveUseAll of these data will be available through published results and in the thesis of the MS student that is on the project. All UI thesis are available electronically through the UI library.
descriptionBiophysical data collected on Moscow Mountain (Latah County, ID) to support the 2019-2020 and 2020-2021 winter camera deployments. CSVs include image metadata, camera coordinates (latitude and longitude in degrees), slope (degrees), aspect (degrees and cardinal directions), canopy cover from densiometer readings taken during camera deployment and derived from hemispherical photography (%), air temperature data from remote camera images and external LogTag (R) TRIX-8 temperature loggers (degrees Celsius), precipitation events (T/F), manual snow depth estimates from "virtual" snow stakes and permanent snow stakes (when applicable; cm), and estimated diffuse and direct shortwave radiation derived from hemispherical photography (W/m^2). Two separate CSVs contain snow density and hardness measurements taken at camera sites throughout the snow-on period approximately monthly.
repositoryNone planned at this time.
qualityChecksLogTags were hung from the same heights as cameras to ensure that temperature differences were not the result of elevational gradients in temperature. Cameras were checked throughout deployment for proper function. Hemispherical photographs were only taken at dawn and dusk or on overcast days to ensure good lighting conditions. The hemispherical camera was carefully levelled for each hemispherical photograph. Settings in Hemisfer were tailored to each image to ensure proper thresholding. Images were edited when necessary to remove an object which was interfering with thresholding.
protocolsLogTag temperature sensors were programmed to take temperature readings every 45 minutes throughout their deployment. LogTags were housed in protective plastic cases and then in home-made radiation shields made of PVC and aluminum foil tape. A LogTag in its housing was hung from a tree at the same height as the camera with which it is associated. Technicians processed images to derive precipitation events and snow depth. See the "Automated Image Database for Moscow Mountain (Latah County, ID)" section of this DMP for further details about camera deployments and image analysis. Shortwave radiation and canopy cover were estimated using hemispherical photography. At each camera site, a Canon 70D camera and Sigma 8mm fisheye lens were used to take hemispherical photographs 1) directly at the tree to which the camera was mounted and 2) 5m into the camera viewshed. Hemispherical photographs were manually thresholded and processed using Hemisfer (R) software, and radiation values were output at half-hour intervals for every day of the year.
format.csv
restrictionsNo restrictions
backupAndStorageAll data have been backed up on an external hard drive.
dataManagementResourcesData will be housed in an OwnCloud directory structure maintained by the Northwest Knowledge Network. $1000 (.33%) of the budget are allocated to cover NKN fees in addition to M.S. student time required for data management tasks and field technician time for conducting fieldwork.
volumeEstimate234 MB
dataProcessingLogTag data were extracted from sensors using LogTag Analyzer (R) software. Images were processed in Timelapse2 (R) software. Hemispherical photographs were processed using Hemisfer (R) software. All data streams were then joined into a single CSV using Program R.
nameBiophysical Database from Moscow Mountain (Latah County, ID)
phaseSubmitted DMP
templateNameCASC DMP v7

Item Actions

View Item as ...

JSON ATOM ISO XML

Save Item as ...

JSON FGDC MODS XML ISO XML

View Item...

Metrics