Model structure & program control parameters -- California Nox -- log 2wk aves #28 N: number of sites #124 T: number of temporal replications in covariances #n V: variance field, constant (“c”) or nonconstant random field (“n”) #e C: covariance model, exponential (“e”) or power exponential (“p”) #24 i1: indices of the two sites to be held fixed in the D-plane #21 i2: " " " #200 I: how many MCMC iterations to run before printing/saving results #2000 J: the number of samples to print/save (total number of iterations is I*J) Spatial deformation parameters #.01 tau: the scale (variance) parameter for the multivariate normal prior for the parameters, W , of the nonlinear component of the thin-plane spline #.05 w_mcs: MCMC sampling s.d. for the spline nonlinear parameters W #100 s1a: prior variances of the two spline linear parameters #100 s2a: #.1 mcsa: MCMC sampling s.d. for the two spline linear parameters D-plane correlation parameters #.3 theta: initial value for parameter theta of the D-plane [power] exponential correlation function [note dependence of this parameter on the range of values in the coordinate system] #1 ptheta: parameter of the exponential prior on theta [note: I don’t see any reason to pick different forms of this prior, uniform or exponential, depending on whether the model is constant or nonconstant variance (according to Doris’ response to my notes on program changes). Let’s stick with one or the other. Most analyses will probably be nonconstant variance. Please advise Gabriel regarding changes to code.] #.15 theta_mcs: MCMC sampling s.d. for theta (sampling is from a gamma distribution with mean being the current MCMC estimate) [to be changed from current inverse variance, currently called theta_toggle. i.e., new parameter theta_mcs is sqrt(1/theta_toggle), as currently defined] #1.5 theta2: initial value for parameter theta2 of the D-plane power exponential correlation function #2 ptheta2: upper limit of the uniform prior on theta2 [this is reasonably left fixed at the value 2 as the domain of theta2 is (0,2).] #.3 theta2_mcs: MCMC sampling s.d. for theta2 (sampling from a gamma distribution with mean being the current MCMC estimates) [to be changed from current inverse variance, called theta2_toggle?] #.001 nugget: initial value of the nugget parameter of the D-plane correlation function #1 pnugget: parameter of the exponential prior on the nugget #.4 nugget_mcs: MCMC sampling s.d. for nugget [to be changed form current inverse variance, called nugget_toggle: nugget_mcs = sqrt(1/nugget_toggle)] Variance field: #0 imu: initial value of the mean of the log variance field [not in current program] #6 isigma2: initial value of the variance of the log variance field [not in current program] #2 pmu: mean of the normal prior for the parameter mu #8 psigma2: variance of the normal prior for the parameter mu #20 palpha parameter alpha of the gamma prior for the variance of the log variance field #5 pbeta parameter beta of the gamma prior for the variance of the log variance field [Note that there are no MCMC sampling parameters because mu and sigma2 are updated using Gibbs steps, rather than Hastings. But there is a Hastings step for the values of the variance field at the monitoring sites.] #98 nu_lb: MCMC uniform proposal distribution for the elements of the variance field at the monitoring sites, nu_i, is [(nu_lb/100)*current value,(100/nu_lb)*current value]. #1 thetat: initial value of the parameter thetat of the exponential spatial correlation function for the variance field #1 pthetat: parameter of the exponential prior on thetat #.35 thetat_mcs: MCMC sampling s.d. for thetat [to be changed from the current prompt for an inverse variance]