Climate Space Model

openModeller id: CSMBS

Current version: 0.4    Developer(s): Tim Sutton, Renato De Giovanni

Accepts Categorical Maps: no

Requires absence points: no

Author(s): Neil Caithness


Climate Space Model [CSM] is a principle components based algorithm developed by Dr. Neil Caithness. The component selection process int this algorithm implementation is based on the Broken-Stick cutoff where any component with an eigenvalue less than (n stddevs above a randomised sample) is discarded. The original CSM was written as series of Matlab functions.


Robertson M.P., Caithness N., Villet M.H. (2001) A PCA-based modelling technique for predicting environmental suitability for organisms from presence records. Diversity and Distributions, 7:15-27


Number of random eigenvalues

openModeller id: Randomisations

The Broken Stick method of selecting the number of components to keep is carried out by randomising the row order of each column in the environmental matrix and then obtaining the eigen value for the randomised matrix. This is repeatedly carried out for the amount of times specified by the user here.

Data type: integer  Domain: [1.0, 1000.0]  Typical value: 8

Number of standard deviations

openModeller id: StandardDeviations

When all the eigen values for the 'shuffled' environmental matrix have been summed this number of standard deviations is added to the mean of the eigen values. Any components whose eigen values are above this threshold are retained.

Data type: real  Domain: [-10.0, 10.0]  Typical value: 2.0

Minimum number of components in model

openModeller id: MinComponents

If not enough components are selected, the model produced will be erroneous or fail. Usually three or more components are acceptable

Data type: integer  Domain: [1.0, 20.0]  Typical value: 1

Show very detailed debugging info

openModeller id: VerboseDebugging

Set this to 1 to show extremely verbose diagnostics. Set this to 0 to disable verbose diagnostics (this is default behaviour).

Data type: integer  Domain: [0.0, 1.0]  Typical value: 0