Skip to contents

Tapered Change-of-Support Spatial Models

cosTapered-package.Rd

The cosTapered package implements a Bayesian change-of-support spatial model for areal observations, fine-support raster covariates, and areal prediction units. The main Gaussian response model uses a tapered exponential covariance. Experimental Polya-Gamma response paths are available for binomial and fixed-size negative-binomial observed-support responses. Observed-support and prediction-support weights are computed from polygon and raster intersections.

Details

The main workflow is:

  1. Build observed-support objects with cos_prepare.

  2. Specify priors with cos_default_priors.

  3. Fit the collapsed posterior with cos_fit.

  4. Recover observed-support latent quantities with cos_recover_B.

  5. Predict fine-support or areal-support quantities with cos_predict_fine and cos_predict_areal.

Users provide covariate rasters as prepared inputs. The package does not add intercepts, scale covariates, or buffer points automatically.

The package vignettes provide the main workflow, model details, and spatial guidance. A model and software details document with compact model notation and posterior calculations is installed as doc/cosTapered.pdf when the package is built with vignettes.

Author

Andrew O. Finley, Lu Zhang, Arne Nothdurft, and Sudipto Banerjee.

References

Zhang, L., Finley, A. O., Nothdurft, A., and Banerjee, S. (2024). Bayesian modeling of incompatible spatial data: A case study involving Post-Adrian storm forest damage assessment. International Journal of Applied Earth Observation and Geoinformation, 135, 104224. doi:10.1016/j.jag.2024.104224