Using openModeller command-line tools on Windows

You have installed the openModeller command-line interface (CLI). openModeller is a modular software and there are different ways to use its core functionality. If you are interested in a graphical user interface (GUI) with buttons, controls and images you should download a different software package called openModeller Desktop. Instead, if you want to learn and use the CLI interface, the package you have just installed offers tools (commands) that provide a similar funcionality to openModeller Desktop, except for the visualization facilities. In this case, you can use a geographical information system, such as Quantum GIS, to visualize your data.

This document is not intended to be a replacement for the manual pages nor for the documentation available in the openModeller website. This is just a quick introduction about the openModeller command line tools including an example of how to generate a simple model.

Getting started

Command line tools run on a terminal. On Windows, the terminal is the Command prompt, also known as the MS-DOS prompt. Starting from openModeller 1.3.0, the installer includes shortcuts in the Start menu to all openModeller commands available in this environment (there are a couple of additional commands that are only available under GNU/Linux). When you click on any of these shortcuts, a new terminal is opened called openModeller Terminal, which is basically a standard Command prompt configured to automatically run the associated command line tool without any parameters, forcing the tool to output usage help. This output is basically the set of parameters required to effectively use the tool.

The openModeller Terminal starts in the Working Folder specified during installation (the default is C:\om). Inside that folder, there is another folder called examples containing a couple of environmental layers, a file with point data and an example of a request file used to generate models.

Generating a model

From now on, the commands you should type to reproduce the example are in bold, while the content in monospaced font shows the output from the terminal itself and/or openModeller. We assume the Working Folder to be the default C:\om.

Click on the om_console shortcut in the Start menu to get into the prompt, and then enter the examples directory:

C:\om> cd examples

To generate a model with om_console you just need to provide a single parameter, which is a request file actually containing all possible parameters for this task. Use the request file provided as example:

C:\om\examples> om_console request.txt
[Info] openModeller version 1.3.0
[Info] Reading occurrences...done
[Info] Reading layers...done

Choose an algorithm between:
[1] AquaMaps (beta version)
[2] Bioclim
[3] Climate Space Model
[4] GARP (single run) - DesktopGARP implementation
[5] GARP with best subsets - DesktopGARP implementation
[6] ENFA (Ecological-Niche Factor Analysis)
[7] Envelope Score
[8] Environmental Distance
[9] GARP (single run) - new openModeller implementation
[10] GARP with Best Subsets - new openModeller implementation
[11] Maximum Entropy
[12] Niche Mosaic
[13] Artificial Neural Network
[14] Random Forests
[15] SVM (Support Vector Machines)
[q] Quit

Option: 2
> Algorithm used: Bioclim

[Info] Uses mean and standard deviation for each environmental variable
separately to calculate bioclimatic envelopes. Level of fitness between
the environmental values on a point and the respective envelopes
classified points as Suitable, Marginal, or Unsuitable for

* Parameter: Standard deviation cutoff

The envelope is determined by multiplying this
parameter and the standard deviation.
Standard deviation cutoff >= 0.000000
Enter with value [0.674]: [Enter]
[Info] Creating model
[Info] Reading 2-dimensional occurrence
[Info] Using 65 points to find the bioclimatic envelope.
[Info] Finished creating model
[Info] Calculating ROC curve
[Info] Generating 10000 background points
[Info] Projecting model
[Info] Finished projecting model
[Info] Model statistics for training data
[Info] Threshold: 50.00%
[Info] Accuracy: 100.00%
[Info] Omission error: 0.00% (0/65)
[Info] Lowest prediction: 0.50
[Info] AUC: 0.72
[Info] Projection statistics
[Info] Threshold: 50%
[Info] Cells predicted present: 66.68%
[Info] Total number of cells: 64681
[Info] Done.

That's it. If you open request.txt at C:\om\examples with a text editor you will see all parameters that were used in the modelling procedure, including the file with point data, the environmental layers used for model creation and model projection and so on. As a result you get the openModeller serialized model in XML (in this case furcata.xml) and the projected model (furcata.img) that you can open in a GIS that supports this format (which is ERDAS Imagine). Note that you can also change that format in the request if you want to produce a GeoTiff or an ARC/Info ASCII Grid instead.

In order to produce models with your own data, you can create a copy of request.txt, edit the new file and then run the same command passing the new file as a parameter.

Additional information

For more information about the openModeller command line tools, please check the online documentation.

To interact with the openModeller community you can subscribe to the users mailing list.

For more information about openModeller in general, you can read the openModeller paper. The same paper can also be used to cite openModeller:

Muņoz, M.E.S., Giovanni, R., Siqueira, M.F., Sutton, T., Brewer, P., Pereira, R.S., Canhos, D.A.L. & Canhos, V.P. (2009) "openModeller: a generic approach to species' potential distribution modelling". GeoInformatica. DOI: 10.1007/s10707-009-0090-7