In many applications, spatial data often display heterogeneous dependence patterns and may be subject to irregular geographic constraints. In light of these challenges, this dissertation develops several novel Bayesian ...