'spup' - An R package for uncertainty propagation in spatial environmental modelling

K. Sawicka*, G.B.M. Heuvelink

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference paper

1 Citation (Scopus)

Abstract

Computer models are crucial tools in engineering and environmental sciences for simulating the behaviour of complex systems. While many models are deterministic, the uncertainty in their predictions needs to be estimated before they are used for decision support. Advances in uncertainty analysis have been paralleled by a growing number of software tools, but none has gained recognition for universal applicability, including case studies with spatial models and spatial model inputs. We develop an R package that facilitates uncertainty propagation analysis in spatial environmental modelling. The 'spup' package includes functions for uncertainty model specification, propagation of uncertainty using Monte Carlo (MC) techniques, and uncertainty visualization functions. Uncertain variables are represented as objects which uncertainty is described by probability distributions. Spatial auto-correlation within a variable and crosscorrelation between variables is also accommodated for. The package has implemented the MC approach with efficient sampling algorithms, i.e. stratified random sampling and Latin hypercube sampling. The MC realizations may be used as an input to the environmental models called from R, or externally. Selected static and interactive visualization methods that are understandable by nonstatisticians can be used to visualize uncertainty about the measured input, model parameters and output of the uncertainty propagation.

Original languageEnglish
Title of host publicationProceedings of Spatial Accuracy 2016
EditorsJean-Stéphanie Bailly, Daniel Griffith, Didier Josselin
PublisherInternational Spatial Accuracy Research Association (ISARA)
Pages275-282
ISBN (Print)9782910545105
Publication statusPublished - 2016
Event12th International Symposium on Spatial Accuracy Assessment in Natural Resources and Environmental Sciences, Accuracy 2016 - Montpellier, France
Duration: 5 Jul 20168 Jul 2016

Conference

Conference12th International Symposium on Spatial Accuracy Assessment in Natural Resources and Environmental Sciences, Accuracy 2016
CountryFrance
CityMontpellier
Period5/07/168/07/16

    Fingerprint

Keywords

  • Monte Carlo
  • R language
  • Spatial models
  • Uncertainty analysis
  • Uncertainty propagation

Cite this

Sawicka, K., & Heuvelink, G. B. M. (2016). 'spup' - An R package for uncertainty propagation in spatial environmental modelling. In J-S. Bailly, D. Griffith, & D. Josselin (Eds.), Proceedings of Spatial Accuracy 2016 (pp. 275-282). International Spatial Accuracy Research Association (ISARA).