Issue with Patcher in LUC simulation

[katherine...@gmail.com]

Hi,

I'm trying to run a model to simulate deforestation (step 5 in the land use change tutorial) and running into a problem with the Patcher functor. I am interested in the transition from land class 1 to 2 (from a land cover map with 5 categories), so I had initially filled in the Patcher with mean patch size, patch size variance, and patch isometry values for that transition only. But when I tried to run the model, I got an error message: "Repeat": Contained "Patcher" (alias "patcher 2712") found an error: Mean, variance and isometry for transition 0->1 are not defined. In the message log, it also showed expected transitions for all the possible transitions between the 5 land cover classes, not just the one transition I'm interested in, although I'm not sure if that is itself an issue. 

To try to avoid this error, I went back to the Patcher and assigned mean patch size, patch size variance, and patch isometry values for all the transitions that are present in the transition matrix and re-ran the model. Now I am getting a different error message: "Repeat": Contained "Patcher" (alias "patcher 2712") found an error: The layer "probability_0_to_1" is not defined. 

So I can't run the model without giving patch size info for the 0->1 transition, but the model can't run because the probability layer for that transition is not defined? Any ideas on how to get out of this error loop?

Thank you!

[Hermann Rodrigues]

Hi,

You can solve the problem: you can ask the Calc W Of E Probability Map to calculate the probability layer or you can remove the transition from your transition matrix. Removing the transition from the transition matrix is usually the right thing to do, since you probability does not want the transition 0->1 to be part of your simulation.

Best,

hermann

----------------------------------------------------------------------
Hermann Rodrigues
her...@csr.ufmg.br
hermann....@gmail.com
Skype: hermann.rodrigues
Twitter: @horodrigues / @dinamica_ego

Centro de Sensoriamento Remoto / UFMG
http://www.csr.ufmg.br / http://www.csr.ufmg.br/dinamica

--
You received this message because you are subscribed to the Google Groups "Dinamica EGO" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dinamica-ego...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dinamica-ego/10ced56c-caa0-427c-818c-01d38261b963%40googlegroups.com.

[katherine...@gmail.com]

I tried your suggestion of modifying the transition matrix and it worked! Thank you for the speedy and helpful response!

To unsubscribe from this group and stop receiving emails from it, send an email to dinami...@googlegroups.com.

[Pablo Benitez]

Hello, I am about to run my LUCC model and it keeps popping up the same mistake in patcher (Error: The layer #1 is not defined). I really don't know what the problem is. 

My data are the following: 

1. Initial cover map: Forest (1)- Not forest (2). 

2. Static variables: Distance euclidiana to rivers, main roads, piscicola, etc 

3. I'm using the multi-step transition matrix.

4. The period of time is from 2003 to 2015.

5. I only have 1 transition, from forest (1) to not forest (2)

Thank you for your help!

[Hermann Rodrigues]

Hi,

Based on the error message, it appears that somehow the transition matrix is informing Patcher that your model is dealing with two (or more) transitions, but there is only one probability layer in the probability map.

How is the probability map being calculated? Are you using the Calc W Of E Probability Map?

And are you loading the transition matrix from a CSV file? Can you confirm that there is only one transition being mentioned in that file?

hermann

----------------------------------------------------------------------
Hermann Rodrigues
her...@csr.ufmg.br
hermann....@gmail.com
Skype: hermann.rodrigues
Twitter: @horodrigues / @dinamica_ego

Centro de Sensoriamento Remoto / UFMG
http://www.csr.ufmg.br / http://www.csr.ufmg.br/dinamica