GIS Analyses

ArcGIS was used for all of the processing steps and some of the analysis steps.  Biomapper was used for most of the analysis steps.  The following are step-by-step protocols for the processing and analysis steps used in this project.

* See Flowchart for further information on the relation and order of the processing and analysis steps.

 

ArcGIS Processing and Analysis Steps

Convert ASCII files to GRID files

We used several data layers (e.g., “npp”, “landcover”) that were originally saved in ASCII format.  Files in ASCII format were not usable in Biomapper (these files were not convertible to RST [IDRISI] format), so they needed converting to GRID files.

1a.) Open ArcGIS (ArcCatalog or ArcMap).  1b.) Click on the “Show/Hide ArcToolbox Window” button on the menu bar.  1c.) In the ArcToolbox table, expand “Conversion Tools”, expand “To Raster” and double-click on “ASCII to Raster”.  2a.) In the ASCII to Raster window, click on the folder button next to the entry line “Input ASCII raster file”: find the ASCII file (e.g., “npp”, “landcover”).  2b.) In the ASCII to Raster window, click on the folder button next to the entry line “Output raster”: either name the new file meaningfully or create a new folder with a meaningful name and save the new file in this new folder.  2c.) In the ASCII to Raster window, click on the arrow button next to the entry line “Output data type (optional)”: choose either INTEGER or FLOAT based on the ASCII file’s properties.  3a.) To find this information, in ArcCatalog, find the ASCII file and right-click on its icon.  3b.) In the menu window, click on the “Properties” button; under “Raster Information”, find “Pixel Type”, which will be either “integer” or “floating point”.  Note: the ASCII file, if it had any projection data, will not have any projection data in the new raster file.  The projection of the new raster file will need to be defined based.

Project GRID files as WGS 1984 UTM Zone 37N

We mostly used data layers (e.g., “bio1”, “precip1”) that were already saved in GRID format.  These files and the new GRID files converted from ASCII format needed projecting into the same coordinate system.

1a.) Open ArcGIS (ArcCatalog or ArcMap).  1b.) Click on the “Show/Hide ArcToolbox Window” button on the menu bar.  1c.) In the ArcToolbox table, expand “Data Management Tools”, expand “Projections and Transformations”, expand “Raster” and double-click on “Project Raster”.  2a.) In the Project Raster window, click on the folder button next to the entry line “Input raster”: find the GRID file.  2b.) In the Project Raster window, click on the folder button next to the entry line “Output raster”: either name the new file meaningfully or create a new folder with a meaningful name and save the new file in this new folder.  2c.) In the Project Raster window, click on the menu button next to the entry line “Output coordinate system”.  3a.) In the “Spatial Reference Properties” window, click on the “Select” button.  3b.) In the “Browse for Coordinate System” window, double-click on the “Projected Coordinate Systems” folder, double-click on the “Utm” folder, double-click on the “Wgs 1984” folder and double-click on “WGS 1984 UTM Zone 37N”.  3c.) In the Spatial Reference Properties window, click on the “OK” button.  4a.) In the Project Raster window, click on the “OK” button.

Resample “npp” GRID file to approximate 1-km cell size

The data layer “npp” had a larger cell size than all of the other data layers, so it needed resampling to a smaller cell size because files with different cell sizes are not usable in Biomapper.

1a.) Open ArcGIS (ArcCatalog or ArcMap).  1b.) Click on the “Show/Hide ArcToolbox Window” button on the menu bar.  1c.) In the ArcToolbox table, expand “Data Management Tools”, expand “Raster” and double-click on “Resample”.  2a.) In the Resample window, click on the folder button next to the entry line “Input Raster”: find the GRID file.  2b.) In the Resample window, click on the folder button next to the entry line “Output Raster”: either name the new file meaningfully or create a new folder with a meaningful name and save the new file in this new folder.  2c.) In the Resample window, click on the folder button next to the entry line “Output Cell Size”: find a GRID file with the exact cell size to which to resample the “npp” GRID file.  2d.) In the Resample window, click on the folder button next to the entry line “Resampling Technique (optional)”: choose NEAREST.  2e.) In the Resample window, click on the “OK” button.

Mask GRID files to “npp” GRID file (“least common denominator” extent)

The data layer “npp” had a smaller extent than all of the other data layers, so all of the other data layers needed masking to the smaller extent because files with different extents are not usable in Biomapper.

1a.) Open ArcGIS (ArcMap), click on the “A new empty map” button and click on the “OK” button.  1b.) Click on the “Tools” button on the menu bar and click on the “Extensions” button.  1c.) In the Extensions window, click on the box next to “Spatial Analyst” and click the “Close” button.  1d.) Click on the “View” button on the menu bar and click on the “Toolbars” button and click on the “Spatial Analyst” button.  2.) Click on the “Add Data” button on the menu bar and add all of the GRID files that need to be masked (i.e., “npp”, “landcover”, “treecover”, “radi”, “elev”, “slope”, “eastness”, “northness”, “bio<1-19>”, “precip<1-12>”, “tmin<1-12>”, “tmean<1-12>”, “tmax<1-12>”).  3a.) In the Spatial Analyst window, click on “Spatial Analyst” and click on the “Options” button.  3b.) In the Options window, click on the “General” tab and click on the arrow next to the entry line for “Analysis mask”: choose “npp”.  3c.) In the Options window, click on the “Extent” tab and click on the arrow next to the entry line for “Analysis extent”: choose “Same as Layer ‘npp’”.  3d.) In the Options window, click on the “Cell Size” tab and click on the arrow next to the entry line for “Analysis cell size”: choose “Same as Layer ‘npp’”.  3e.) In the Options window, click on the “OK” button.  4a.) In the Spatial Analyst window, click on Spatial Analyst and click on the “Raster Calculator” button.  4b.) In the Raster Calculator window, under “Layers”, double-click on any of the other GRID files (e.g., “landcover) and click on the “Evaluate” button.  5a.) A new file “Calculation” is created and added to the map as a layer.  5b.) Right-click on the Calculation layer.  5c.) In the menu window, click on the “Data” button and click on the “Export Data” button.  5c.) In the “Export Raster Data - Calculation” window, click on the folder button next to the entry line “Location”: create a new folder with a meaningful name.  5d.) In the Export Raster Data - Calculation window, click on the arrow button next to the entry line “Format”: choose “GRID”.  5e.) In the Export Raster Data - Calculation window, on the entry line “Name”, name the new file meaningfully and click on the “Save” button.  6.) Repeat steps 4a through 5e for all of the other GRID files (e.g., “treecover”, etc.), which will have the same extent as the “npp” file, the “least common denominator” extent.

Reclassify “landcover” GRID files for each species

The data layer “landcover” was saved as nominal data and needed converting to ordinal data because ecogeographic variable data saved as nominal data are not usable in Biomapper.

1a.) Open ArcGIS (ArcMap).  1b.) Click on the “Tools” button on the menu bar and click on the “Extensions” button.  1c.) In the Extensions window, click on the box next to “Spatial Analyst” and click the “Close” button.  1d.) Click on the “Show/Hide ArcToolbox Window” button on the menu bar.  1e.) In the ArcToolbox table, expand “Spatial Analyst Tools”, expand “Extraction” and double-click on “Extract Values to Points”.  2.) Click on the “Add Data” button on the menu bar and add “landcover” and all of the species shapefiles (i.e., “nyala”, “eland”, “grtkudu”, “leskudu”, “combbk” and “menbbk”).  3a.) In the Extract Values to Points window, click on the folder button next to the entry line “Input point features”: choose one of the species shapefiles (e.g., “nyala”).  3b.) In the Extract Values to Points window, click on the folder button next to the entry line “Input raster”: choose “landcover”.  3c.) In the Extract Values to Points window, click on the folder button next to the entry line “Output point features”: either name the new file meaningfully or create a new folder with a meaningful name and save the new file in this new folder.  3d.) In the Extract Values to Points window, click on the button next to the line “Append all the input raster attributes to the output point features” and click on the “OK” button.  4a.) A new file “Your new name” is created and added to the map as a layer.  4b.) Right-click on the new layer.  4c.) In the menu window, click on the “Open attribute table” button.  4d.) In the “Attributes of ...” window, right-click on the “RASTERVALU” column heading.  4e.) In the menu window, click on the “Summarize” button.  4f.) In the Summarize window, click on the folder button next to the entry line “Specify output table”: either name the new file meaningfully or create a new folder with a meaningful name and save the new file in this new folder.  4g.) In the Summarize window, click on the “OK” button.  5a.) A new file “Your new name” is created and added to the map as a table.  5b.) Right-click on the new table.  5c.) In the menu window, click the “Open” button.  5d.) In the “Attributes of ...” window, right-click on the “Count_RASTERVALU” column heading.  5e.) In the menu window, click on the “Sort Descending” button.  6a.) In the Spatial Analyst window, click on the “Reclassify” button.  6b.) In the Reclassify window, under “New Values” click on the row of the land cover class that had the lowest count in the new table and type “1”.  6c.) Repeat step 6b for all of the other land cover classes, typing “2” through “11” in the order of lowest to highest counts in the new table, which will be ranking all of the land cover classes by the count of species points in each land cover class.  7.) Repeat steps 3a through 6c for all of the other species shapefiles (e.g., “eland”).

This .pdf file includes many of the following processing and analysis steps with relative screen captures.

Add new field to attribute tables

Add new field to species shapefiles and code uniformly to some number (e.g., “2”).  Create a new field in the attribute table (e.g., “new_values”).  Make sure this is a different value from the background value of the study area shapefile with which the species file has to be merged.  Use the “same as ...” option.

Convert shapefiles to GRID files

Convert shapefile to raster (use “conversion tools”).  Make sure you select the newly added field with a constant value for all animal locations (e.g., “2”) in the field ID.  In the environments, go to “general” and set the extent to the same as the study area raster.  Under raster analysis settings, set the cell size to that of the study area raster.

Mosaic species GRID files and Ethiopia GRID file

Mosaic to new rasters (use “data management tools”; alternatively, the raster calculator in spatial analyst can also be used after standardizing the extent and cell size).  Put in the study area raster and the species raster > mosaic.  Once again, you need to set the extent and cell size to that of the study area raster in the environments.  In the list of added rasters that are to be merged, keep your species raster below the study area raster.

Reclassify GRID files as Boolean data

Reclassify in spatial analyst to make sure the background (beyond the study area boundary) is assigned “no data”.  The background within the study area should be assigned “0”.  The species locations should be assigned “1”.  This is a Boolean map.

Convert GRID files to RST files (IDRISI)

Convert rasters to IDRISI format.  The plug-in is often quirky.  It usually works best if the binary data species raster is rendered in “stretched” mode (use “properties – symbology”) before the GRID > IDRISI conversion is carried out.  The GRID > IDRISI converter has plug-ins for ArcMap and ArcView.  Make sure all files are saved in a hard drive (internal or external) but not on the desktop or my documents for this converter to work.  If at first the conversion does not work, retry several times; it usually eventually works.  Sometimes it will appear as though the output RST (IDRISI) file is not Boolean.  Go into properties – symbology in ArcMap and select “classified”. Click ok. Go back into properties and select “stretched”.  Your RST file may appear as a Boolean file now.

Things to consider in preparing Boolean (or binary) files for species data

– Make sure the study area raster (or shapefile) and the species files are in the same datum and projection.
– You may need to convert the study area shapefile into a raster (use “feature to raster”).  It may make sense to set the environments to those of some other file you will use in the analysis from the EGV layers (e.g., the DEM).
– Make sure you name each new file created using some sensible nomenclature scheme – saving each set of files in clearly labeled folders helps.
– Make sure you have all master files backed up in a directory separate from your working directory.
– Make sure the coordinate data for species presence is correctly entered.

(return to top of page)

 

Spatial Analysis

Without spatial analysis, geographic information systems are not much more than a new and perhaps improved mode of cartography.  However, geographic data include attributes and locations, which are very important for solving problems in landscape ecology but are dependent on the scale and resolution of the data.  The scale of many ecological studies is too small to scale up to landscape dynamics, whereas the resolution of many data, such as remotely sensed data, is too coarse to resolve down to management scenarios.  However, there seems to be a relatively happy medium between large- and small-scale and coarse- and fine-resolution of data available for use in species distribution models of many organisms.

SPATIAL ANALYSIS – The process of examining the locations, attributes, and relationships of features in spatial data through overlay and other analytical techniques in order to address a question or gain useful knowledge.

RESOLUTION – The detail with which a map depicts the location and shape of geographic features.  The larger the map scale, the higher the possible resolution.

We used ecogeographical variable data that had resolution to approximately 1 square kilometer, except for one data layer (“npp”), which was resolved to approximately 81 square kilometers.

SCALE – The ratio or relationship between a distance or area on a map and the corresponding distance or area on the ground, commonly expressed as a fraction or ratio.

We used species data and ecogeographical variable data that were scaled to the entire country of Ethiopia.

CELL SIZE – The dimensions on the ground of a single cell in a raster, measured in map units.  Cell size is often used synonymously with pixel size.

EXTENT – The geographic bounding area within which spatial analysis will occur.  The bounding area is set by defining the x,y coordinates of opposite corners, usually the bottom-left and top-right corners of results.

MASK – In ArcGIS, a means of identifying areas to be included in analysis.

We masked all of the predictor layers for use in Biomapper to the “least common denominator” data layer (“npp”).

MOSAIC – A raster dataset composed of two or more merged raster datasets.

We mosaiced each of the species layers to the Ethiopia layer.

BOOLEAN EXPRESSION – An expression, named for the English mathematician George Boole, that results in a true or false (logical) condition.

For example, we used a Boolean expression to give locations of species a value of 1, or true, and the background of Ethiopia a value of 0, or false.  Also, cells outside the boundary of Ethiopia were given a value of NoData.

RASTERIZATION – The conversion of points, lines, and polygons into cell data.

We converted the vector data on locations of species to raster data.

RECLASSIFICATION – The process of taking input cell values and replacing them with new output cell values.

We changed the land cover for each species from nominal data to ordinal data.

RESAMPLING – The process of interpolating new cell values when transforming rasters to a new coordinate space or cell size.

We resampled one data layer (“npp”) from approximately 9 km by 9 km cell size to approximately 1 km by 1 km.

RASTER CALCULATOR – An ArcGIS Spatial Analyst tool for performing mathematical calculations with operators and functions, setting up selection queries, or typing Map Algebra syntax.

We used raster calculator to mask the 6 species layers and the 78 ecogeographic variable layers to be guarantee the cell size and extent of each layer was equal and would overlay exactly.

NODATA – In raster data, the absence of a recorded value.  NoData does not equate to a zero value.

Because raster data have a rectangular extent and Ethiopia is an irregular shape, many cells do not have values.  To spatially analyze raster data, whether with species distribution models or not, cells with no value instead need NoData values.  If cells did not have NoData values, the predictor layers would not have been usable in Biomapper because it would not have been able to distinguish between cells of Ethiopia and cells of the background outside Ethiopia.

ALGORITHM – A mathematical procedure used to solve problems with a series of steps.  Algorithms are usually encoded as a sequence of computer commands.

Biomapper uses an algorithm to perform ecological niche factor analysis and to create information about habitat suitability.  It uses somewhat of a “black box” approach because it is unknown how the algorithms work.

OPTIMIZATION – The process of fine-tuning data, software, or processes to increase efficiency, improve performance, and produce the best possible results.

Biomapper, like other software used for species distribution modeling, is a program that creates information about habitat suitability, which should be optimal if the predictor variables were appropriate for modeling the suitable habitat for a species based on its ecological requirements and if the data had satisfactory accuracy and precision.

Source: Environmental Systems Research Institute. 2006. GIS Dictionary. Environmental Systems Research Institute Inc., Redlands, CA. http://support.esri.com/index.cfm?fa=knowledgebase.gisDictionary.gateway.

* All definitions (shown in blue) are provided verbatim from the GIS Dictionary on ESRI’s website.

(return to top of page)