Skip to content

SDM Notes

Carlos Gabriel Asato edited this page Jul 18, 2019 · 4 revisions

Notes taken from the original ArcSDM.

TOC Fuzzyfication

This fuzzification method utilized the symbolization of the input raster that has been applied in the map document table of contects (TOC). The symbolization in the TOC defines the number of classes and this tool rescales those classes (1...N) to the range [0,1] by (C - 1)/(N-1) where C is the class value and N is the number of classes.

INPUT

Input Raster Reclass Field Number of Classes

OUTPUT

Fuzzy Membership Raster

Calculate Bends

Convert a polyline feature class to a points feature class whilst calculating the degrees of turning at each point for the interior vertices of the polyline. Right bend is negative; left bend is positive. Bend=400 signifies first and last vertices of the polyline.

INPUT: LineString

OUTPUT: Point

Pseudocode

Characterize Polylines

Calculate various statistics that define the character of a polyline from length based subsets of the entire line. For example, can be used to find subtle bends within a longer curving polyline. See the Tool Help for definitions of the attributes that will be written to the output feature class. See Usage Tips in the Tool Help file for an example analyses

INPUT: LineString

targetLength

MinimunAngle

MaximunAngle

OUPUT: MultiLineString? (polyline)

Pseudo code

Check Data Areas

This tool reports the area of the input Evidence rasters. Use this tool to confirm that the evidence rastes is properly cliipped to the Study Area. Note that slight differences in areas can occur due to differences in cell size or origin. Small differences, less than a few percent, can be ignored. An evidence raster with considerably larger area than the study area can distort the weights considerably. The Environment parameters for an SDM model should all be set before running this tool. See the Help for the Calculate Weights tool for details on the Environment parameters.

Contact Proximity

Creates a proximity raster for proximity to the selected contacts or boundaries between map classes in a raster. This raster is an integer raster of the distance to a selected contact for use in any of the modeling methods. This tool is typically used with categorical rasters.

#" Feature Directional Proximity

This tool creates a raster for addressing questions having to do with directional proximity to a selected set of points, lines, or polygons shapefile feature. Directional proximity means that the proximity is only kept in the direction indicated. This tool allows the user to select a subset of the features from a shapefile and then to calculate the distance from the selected features. If directoinal proximity is desired around oriented features such as lines or ellipical polygons, various artifacts can result with this tool. See the Useage Tips for ways to address these artifacts. ALWAYS INSPECT THE RESULTS TO VERIFY THAT THE DESIRED RESULT HAS BEEN PRODUCED.

Input Feature

Input the feature to measure proximity to.

Selection Expression (optional)

Enter an expression to select the desired lines. For more complex selections, do the selection with various selection tools and save the selection as an attribute, for example the attribute might be selected, with values of 0 and 1, where 1 is the selected subset.

Direction Expressoin (optional)

Enter an expression to select the desired direction. Note too narrow of a directional window will cause a artifacts, such as checker board pattern, of selected and not selected near the feature. This problem is mostly eliminated by selecting a broader directional window. Some experimentation may be necessary to get an acceptable result from the directional expression.

Direction is measured counter clockwise from 0 to 360 starting at east. So east is 0, north is 90, west is 180, etc.

Distance Scale Divisor

Enter a number that will be divided into the distance to change the value to a distance consistent with the resolution of the evidence. See the Summary in the help file for more discussion of this issue.

Dir_Prox

Enter the name of the output directional proximity raster. An output value of -1 indicates the cell is outside the directional window. For analyses, it might be useful to reclassify the proximity values so -1 is larger than any value; then cumulative ascending weights can easily be calculated.

Feature Proximity

This tool creates a raster for addressing questions having to do with proximity to a selected set of points, lines, or polygons feature from a shapefile. This tool allows the user to select a subset of the features from a shapefile and then to calculate the distance from the selected features, where distance is rescaled into some appropriate distance units.

Rank Difference

This tool is used to identify where two models have classes that are the same and different. To map these difference, this tool creates a integer raster of the ranked difference betwen two response themes. The tool defaults to the classification used in the TOC for the inputs and assumes that both response rasters have the same number of classes.

Raster Proximity

This tool creates a raster for addressing questions having to do with proximity to a selected set of points, lines, or polygons feature from a shapefile. This tool allows the user to select a subset of the features from a shapefile and then to calculate the distance from the selected features, where distance is rescaled into some appropriate distance units.

Logistic Regression

This tool is a useful complement to Weights-of-Evidence Calculate Response tool as Logistic Regression does not make the assumption of conditional independence of the evidence with regards to the training sites. Using the evidence and assocaited weights tables, this tool creates the outputs the response and standard deviation rasters. The calculations are based on the Gen_Class attribute in the weights table and the type of evidence.

Training Sites Reduction

Reduce number of training sites by random subset of sites or by thinning to one cell per unit areas. This supports the Weights of Evidence assumption that there is one training point per unit cell. Sites are first selected to be within the current mask, then selected by the selected methods.

The Map Display must be refreshed after running this tool in order to display the selected set.