Wildlife Habitat Models and Accuracy Assessment
NR505 – GIS in Wildlife Sciences
Bighorn sheep were historically native to the canyon lands and mountains of northeastern Oregon and western Idaho. European settlement in the early 1900’s negatively impacted the bighorn populations. Hunting, competition with livestock, and parasites and diseases are factors contributing to the elimination of bighorns in Hells Canyon. In the 1970’s bighorn sheep were reintroduced to these canyon lands with varying success. The herds have been monitored by Idaho Fish&Game among others.
More information: http://www.fs.fed.us/hellscanyon/life_and_the_land/wildlife/bighorn-sheep.shtml
In 2002 the Idaho Cooperative Fish and Wildlife Research Unit developed a habitat model describing lambing habitat for bighorn sheep. The input habitat variables to the model are topography, distance to water, land cover type and patch size. The criteria are as follows:
45-315 degrees aspect
31-85 degrees slope
<=1000m from streams
>=2 ha (20,000 m2)
NLCD (National Landcover Habitat Codes) = 12,31,33,51,71
code covertype
11 Open Water
12 Ice/Snow
21 Low Intensity Residential
22 High Intensity Residential
23 Commercial/Industrial/Transportation
31 Bare Rock/Sand/Clay
32 Quarries/Strip Mines/Gravel Pits
33 Transitional
41 Deciduous Forest
42 Evergreen Forest
43 Mixed Forest
51 Shrubland
61 Orchards/Vineyards/Other
71 Grasslands/Herbaceous
81 Pasture/Hay
82 Row Crops
83 Small Grains
84 Fallow
85 Urban/Recreational Grasses
91 Woody Wetlands
92 Emergent Herbaceous Wetlands
Objectives: In this exercise we will develop a spatial model for bighorn sheep lambing habitat, perform an accuracy assessment and learn how such models can automated and run in ArcInfo AML (arc macro language) code. The developed model is a spatial binary model where each pixel is either a 0 or 1 (habitat or non-habitat) within the Redbird herd on Craig Mountain.
Step 1: Open a new ArcMap project and add the following data:
Herd_nlcd National Landcover Raster - 30 m pixels
Herd_dem Digital elevation model Raster – 30 m pixels
Strms Streams scale 1:100000
Step 2: Derive aspect and slope from the DEM. First set the analysis properties for the analysis, i.e. define the working directory, analysis extent and pixel size. You do this in the Spatial Analyst menu under Options.
· Set the working directory to c:\nr505\data\bhs_data (or any other folder where you want your resulting grids to be stored).
· Set the extent to ‘same as Herd_dem’
· Set the Cell size to 30
Select Surface Analysis and then ‘aspect’ in the Spatial Analyst menu. The input surface is herd_DEM, the cell size is 30 and then name the output raster ‘herd_aspect’. The aspect grid is a grid where each pixel represents the azimuth at that particular location. The azimuths are expressed as decimal numbers and the grid does therefore not have a table – you cannot view the attributes.
Similarly, derive slope in degrees.
Step 3: In this step you will derive a new grid ‘distance to streams’ based on the stream vector data. Set the Analysis mask under Spatial analyst Options to ‘herd_DEM’. This limits the ‘distance to streams’ to the area of interest.
· Under Spatial Analyst – Options – General, set the analysis mask to the ‘herd_DEM’ layer.
· Go to Spatial Analyst – Distance – Straigh line distance
Step 4: You are now ready to query for suitable lambing habitat for bighorn sheep. We will use the Raster Calculator tool in Spatial Analysis drop-down menu to find areas vegetated with NLCD land cover types 12,31,33,51,71 , less than 1000 m from a stream, 45-315 degrees aspect and on 31-85 degrees slope.
You can type in the entire query and once (big chance of making a mistake) or query for each variable at a time (the sensible option).
Similarly, create a query for aspect, slope, and distance to streams.
Finally, query for areas where all queries are equal to one (all criteria are fulfilled).
Accuracy Assessment
Step 5. Add the known lambing locations to the habitat map (rb_lamb). Next you would like to find out if the points are located in the predicted habitat (1) or non-habitat (0). This is a point – grid overlay analysis. You can do this as follows:
· Select ArcToolbox – Spatial Analyt Tools – Extraction – Sample
· The name of the input raster is your habitat model
· The input locations is the rb_lamb point data
· Name the output table
The output table contains the x and y coordinates for the points and the value 1 or 0 that the points fall in. Count the number of 1’s and 0’s and compute the model accuracy.
1