Areal downscaling (spatial disaggregation)

General
7

Oh alright, that explains why the summation was confusing to me.

This code is from another post, with different purpose, but if a simple example of a latent variable will help, this may be a starting point. Linear regression with data-dependent error variances - #6 by cmcd

Does that help at all?

It sounds like the difficult part is building the data itself, and probably the indexing inside the model will be a little challenging.

data {
  int n;
  vector[n] z; // observations or estimates
  vector[n] se; // standard errors of the observations
  vector[n] x;  // covariate
}

parameters {
  vector[n] y;
  real alpha;
  real beta;
  real<lower=0> sigma;
}

model {
 // data model
  z ~ normal(y, se);   
 // process model
  y ~ normal(alpha + x * beta, sigma);   
 // parameter models 
    // put priors on your parameters
}
1 Like