I fit the model to southern App data. I uploaded a folder called "southern_Apps" that contain model and call statement. This model includes all suggestions made thus far:
Detection: 3-pass process modified based on Evan's comment
Temporal trend varies among sites (Ty's comment)
Include random-walk process so that variance among sites differ by year (Ty's comment)
I did not upload the MCMC output because we had agreed during our last call that it would not be a good practice to upload .RData file containing model output. I also could not locate RData files in the folder when I tried to commit.
Anyhow, I have noticed that white nose ('wn' in the model statement) does not seem to converge well. I have narrowed the wn variance parameter from dunif(0,100) to dunif(0,5) and ran a moderately long iteration (20,000 burn-in and 10,000 sampling), but values of wn are all over the specified bound of prior values.
sd.wn ~ dunif(0,5) # first order random walk model variance parameter
I
I fit the model to southern App data. I uploaded a folder called "southern_Apps" that contain model and call statement. This model includes all suggestions made thus far:
Detection: 3-pass process modified based on Evan's comment
Temporal trend varies among sites (Ty's comment)
Include random-walk process so that variance among sites differ by year (Ty's comment)
I did not upload the MCMC output because we had agreed during our last call that it would not be a good practice to upload .RData file containing model output. I also could not locate RData files in the folder when I tried to commit.
Anyhow, I have noticed that white nose ('wn' in the model statement) does not seem to converge well. I have narrowed the wn variance parameter from dunif(0,100) to dunif(0,5) and ran a moderately long iteration (20,000 burn-in and 10,000 sampling), but values of wn are all over the specified bound of prior values.
sd.wn ~ dunif(0,5) # first order random walk model variance parameter
I