Predicting Sheep Resource Selection

Ryan Shields
Ryan Shields

September 06, 2022

Predicting Sheep Resource Selection
Predicting Sheep Resource Selection

Study Area

1,012 total hectares near Scofield Reservoir (39.91° -111.16° N, 109 39.88°, -111.12° E) in Utah County, Utah, USA. Elevation ranging from 2,191 to 2,550m.

Data Collection

Six GPS collars with a modified i-gotU GT-600 GPS unit were placed on six ewes from a herd of 600 adult sheep, yielding 20,958 GPS data points.

Generated 20,958 additional random points to ensure adequate characterization of the study area. 

Environmental data (i.e. elevation, water location) came from the Utah Automated Geographic Reference Center. 

Vegetation data were collected from landfire raster dataset. 

Used Society of American Foresters-Society for Range Management (SAF-SRM) cover type we grouped vegetation into 20 different groups based on dominant vegetation type. 

ArcGIS Pro Model Builder Steps

  • Generated elevation data derivatives: slope, aspect, and ruggedness

  • Reclassified aspect into seven directions: north, northeast, northwest, south, southeast, southwest, and west. 

  • Sampled all raster data layers at 10m intervals by creating point layer from DEM cell centers to use in the Extract MultiValues to Points tool

Best model selected based on AIC score.

Variable coefficients from the selected model were used with the variable values from the 10m sample grid (n=110,615) and yielded an RSF prediction score for each 10m cell.

Rasterized point layer and classified by prediction score.

Demo

Tools used

ArcGIS Proi-gotU GT-600 GPS unitModelbuilderR Studio

Plug-ins used

carData Managementlme4MuMInSpatial analyst

tags

DEMExtract MultivaluesGPSRaster CalculatorRaster Conversion

You might also like

Join the community!

We're a place where geospatial professionals showcase their works and discover opportunities.