View All Pages

In-Class Exercise 6 [Locational Model]

Developing Predictive Locational Model for Callalli Area

Variables:

Elevation, Slope, Aspect, Distance to perennial stream.

[SHORTCUT FOR CLASS SRTM_DERIVED layers Links to an external site., Download unzip and add to project]

  • Generate a Slope raster from the Callalli DEM using Degrees and a Z factor of 1

ArcToolbox > Spatial Analyst > Surface > Slope

  • Generate an Aspect raster from the Callalli DEM

ArcToolbox > Spatial Analyst > Surface > Aspect

  • Distance from perennial water using Hydro100k and classes 1-4 as “perennial” and 5-6 as “intermittent”
  • Open Hydro100k table
  • Sort on Size_ column and select 1-4 range.

ArcToolbox > Spatial Analyst > Distance > Euclidean Distance

Output cell size 30

Query against All_ArchID using only Site locations

Open AllArchID Properties and choose Definition Query tab

  • Use the following

All_ArchID.filetype = 'site_a' OR All_ArchID.filetype = 'site_p'

  • Close Properties
  • Series of Zonal Queries as Table using Stats Type: Mean

ArcToolbox > Spatial Analyst > Zonal > Zonal Query with Table

ZonalStats.JPG

  • AllArchID (point locations) to:

-          Elevation

-          Slope

-          Aspect

-          Hydro Dist

  • Output each table to a table that has an abbreviated name (Elev, Slope, Aspect, H20Dist)
    Four tables are added to your project.

Move data into existing Point feature

  • Open All ArchID and Create four Float type fields:

-          Elev

-          Slope

-          Aspect

-          DistH20

  • One at time: Join on ArchID, Keep all records and copy value over using Field Calculator. Break join, proceed to next for all Four Tables.
  • Remove joins. Save All Archid to new Feature in the Callalli GDB called AllArchID_LocModel

Explore Values with GeoStatistical Analyst Histograms

  • Customize menu > Extensions. Check Geosatistical Analyst
  • Customize > Toolbars > Geosatistical Analyst . Dock this toolbar
  • GeoStats > Explore data… Histogram

Make a histogram of the Slope field in AllArchID_LocModel

Open Attribute Table in AllArchID_LocModel as well.

Note that Selections in Histogram also result in selections on map and in attribute table. Change to Elevation field and note the selection also persists across Field Variable selection. Good for exploring patterns!

Where does the Histogram nearly flatten out? Choose a bar around 16-17 deg slope (or 1.6 or 1.7 x 10^-1) and everything left of it. Look in the attribute table – how many records are selected?

With the Definition Query from before (site_a and site_p only selected) you should have 104 total locations and about 80 selected. Therefore nearly 80% of the site locations are found on slopes of under 17 degrees.

Repeat with DistH20.

Use a threashold of about 830m from “streams < 5” as (defined in the Euclidean Distance calculation before), or 0.83 x 10^-3 in the Geostats Analyst Histogram window.

This again should amount to about 80% of the locations though a slightly different selection.

Raster Math

  • Add the Colca_S1 (SRTM 1-arc-sec) DEM. We’ll use this one for the rest of the exercise.
  • Open Spatial Analyst > Map Algebra > Raster Calculator and create the following intersection

( “Slope” < 17 ) & (“disth20” < 830 )

Look at the expression -- what this is doing?

View the output. Make the 0 value transparent color so it looks something like this.

fileRasterCalc.pngRasterCalc.png

Open the Attribute Table for the new Raster and the count of the 1 values shows you the number of 30x30m cells that fall inside the intersection you’ve created.

 RasterCalc2.png

 

Save your result to a PDF. This is one of the assignments for this exercise.

 

II. Predictive Analyst tool

A DEM from the newly available 1-arc-sec SRTM data has been projected to UTM (zone 18 south) and is available for the area. Let’s repeat these steps using

With this understanding of the underlying logic of a simple inductive predictive locational model let’s have a look at a tool that ESRI has created for ArcGIS specificially for this purpose.

The ArcGIS Predictive Analysis Tools are described here Links to an external site.. I have saved a copy for the class to download Links to an external site. (for Arcmap v10.2 but seems to work with 10.3 for our purposes). Download the file, save to the Downloads folder. Unzip the file and then double-click the AddIn to install the toolbar (no need to quit Arcmap). Note that there is a PDF “Getting started with Predictive Analyst” included in the zip as well.

Turn on under Customize menu > Predictive Analysis. Dock the toolbar that appears. It can take a minute to install. If there are issues check under Customize > Add-In Manager.

Steps

Ensure there is no selection (or select a subset of All_Archid for training on a site type)

  • Click the Query Factor Impact Table tool (QFIT)

Open_QFIT.png

  • Select appropriate Input Points layer and Add all four rasters as factors

Query_Exp_editor.png

  • Click Analyze
  • Select Slope and change to Manual and then slide right side of green graph selector until Max: 18 (point where histogram declines)QFIT_Analysis SlopeMax.png
  • Click View Query
  • Click the 10 Weight icon on right to add weighting to the title bar of each clause in the expressions
  • Move the Slope clause to the top and give it a weight of 2
  • Make the Distance from Water a weight of 2 as well.
  • Click Run

In the result close the windows and name it “SlopeX2_Disth20X2” (or something else meaningful) and move the All ArchID layer up so you can see the site locations.

SPOT

Run the SPOT tool

SPOT_window.png

Click in 4 places include so areas with green, red and yellow backgrounds.

Click the eye query and view the weights for each location.

It is an iterative process to come up with appropriate ranges for each source raster and exploring the resulting output.

Right click the Analysis result and choose Open in Query Editor to return and tweak the inputs. Click Run and revisit the Spot tool.

 

Save a PDF of the result. This is your second file for this assignment.