1

suitability maps in CA_MARKOV

Jungers Quentin

June 20, 2012 16:20

Answered

Follow

Dear all,

After reading some articles about simulation of land-use changes with the CA_MARKOV module I still have a doubt about the suitability maps (also called transition potentiel maps). Do the high values for a cell in these maps mean a high probability to find a particular land use in this cell at the end of the simulation or do they mean a high probability to be convert in an other land use type? How do the CA_MARKOV module interpret these maps as input?

Thanking you in advance for answering.

Best regards,

 

Quentin

• Facebook
• Twitter
• LinkedIn

11 comments

• 0
  
  Gabriela Morales July 29, 2012 23:55

  Quientin i am sorry i cannot awnser that. Maybe it says something in the Help menu of IDRISI. On other subject, how do you create suitability maps on the first place? did you do it with the MCE Module?

   

  Cheers!

• 0
  
  Clark Labs August 01, 2012 18:43

  Quentin,

  The transition potential maps, or suitability maps, could or could not be probabilities, depending on how you created them. But think of them as suitability scores ranging from 0 to 255, with 255 being the highest suitability. How CA_Markov interprets these values can be found in the Help for the MOLA module. 

  Given the limitations of CA_Markov, you may want to consider LCM for your simulation. You will not have to create the transition potential maps before hand, these are created for you from your driver variables. 

  James Toledano

  Clark Labs

• 0
  
  Jungers Quentin August 23, 2012 14:28

  Thank you for the answer!

  I got another question! It is said that Markov processes are stochastics because they give probabilities of transition beetween land uses on the basis of observed changes. But, when I do several comparaison with the MARKOV module I always get the same probabilities of transition. So, Is there a stochastic component in the CA_MARKOV module for land use change prediction and if ther is one, where does it come from?

  Thank you very much.

  Best regards,

  Quentin

• 0
  
  Gabriela Morales October 29, 2012 07:44

  Dear James Toledano:

  I have a question that has to do with what you mention, that LCM is better for simulation than Markov-CA. The IDRISI manual says LCM uses Markov chains for change prediction...so my question is: does it use Markov only or does it incorporate cellular automata as well? It doesn't specify but i am guessing it incorporates CA otherwise how can it allocate the changes? (Markov alone doesn't account for neighborhood). And even if it uses Markov Chains-CA, why is it any different than building a Markov model independently?

   

  Hopefully my question makes sense. Kind regards,

   

  Gabriela.

• 0
  
  Sayem Zaman January 29, 2013 17:58

  There is a lot discussion going on about Markov/CA_Markov. I have a question for you guys, I'll appreciate any answer.

  When i was running Markov module, it creates separate Land cover classes being on that class as a Conditional probability images. How can i make a complete image/map which represent the all classes together in one map in the projected date?

  I know there is Breakout module, which can be used to create Boolean images. Is there any module which can interpret several(5 images/classes) in one map/image?

  Any help is really appreciable.

• 0
  
  Clark Labs Tech Support Team February 01, 2013 15:45

  Dear Sayem,

  The Reclass module can be used to aggregate classes in an image.

   

  Best regards,

  Andrew Hostetler

  Clark Labs Tech Support Team

• 0
  
  Asri Lionel Renggo May 23, 2013 16:37

  hi.. dear all

  I'm working my thesis about coral reef CA_Markov Chain and i have five class land cover (Live coral, dead Coral, rubble, sand, and seagrass), but I'm still confuse for driving factor and constrain for MCE, because classes condition on the sea and land very different. like rubble and dead coral it's can't create suitability map but this class can be change become coral reef class. can you give me suggestion or opinion?

  best regard

  agus aris

     

• 0
  
  Ingrideut July 24, 2016 14:10

   

  Hello
  I need help.
  When I'm making the prediction Chande demand modeling ,
  and I click Run, I get the following message:
  Transition potencial file "LCM" Transition Potencial from 1 to 3.rst
  No found. Plase make sure than transition potential image was created for each transition
  being modeled. 
  Please help me. 
  Thank you! 
  Ingrid
  
  

  Edited by Ingrideut July 24, 2016 14:10
• 0
  
  o2 August 09, 2016 22:49

  Hello everyone,

  
  Now I have a serious doubt concerning CA_MARKOV. What means exactly “transition suitability image collection”? Is not the same a “transition map” that a “suitability map”? . The Terrset documentation says “Suitability maps for each land cover (typically produced using MCE and supplied by the user) establish the inherent suitability of each pixel for each land cover type. These are no “transition map”. Why is named “transition suitability image collection” in CA_MARKOV module?

  Thanks. 

   

• 0
  
  o2 August 09, 2016 23:00

  In the same way, What map we choose for calculate the suitability for each classes? ¿Initial-state, final-state or persistence-state? 

• 0
  
  Denisa Lavinia Badiu February 12, 2018 11:26

  Greetings,

  I am using TerrSet to make change predictions using the LCM model, Change Prediction sub-models. I started with the first sub-model: Change analysis. Is it necessary to also go through the Transition Potentials sub-model in order to move to the Change Prediction sub-model or is it ok to just do the two of them: Change analysis and Change Prediction.

   

  Thank you,

  Denisa