Scaling computational costs with data size

Characterize how the computational costs of Bayesian nonparametric ordinal drought models with spatial dependence scale as the amount of data increases.

Background

The paper emphasizes that Bayesian nonparametric priors already increase computational cost and model complexity, and that adding spatial dependence would increase them further. The authors leave unresolved how these costs behave as the number of observations, locations, or other data dimensions grows.

This problem is distinct from merely developing a spatially dependent model: it asks for an explicit characterization of the computational scaling of the resulting methodology.