Input data for environmental models may have been measured in the field or laboratory, derived from remotely sensed imagery or obtained from expert elicitation. Data are also often digitised, interpolated, classified or generalised prior to submission to a model. In all these cases errors are introduced, which gives rise to uncertainty about the ‘true’ value of model inputs. Although users may be aware that errors and uncertainty propagate through their models, they rarely pay attention to this problem. However, when the accuracy of the data is insufficient for the intended use then this may result in poor model results, wrong conclusions and bad decisions.
In this session we introduce some of the concepts that underline uncertainty propagation analysis in spatial environmental modelling. The emphasis will be on Monte Carlo simulation methods as implemented within the ‘spup’ package in R. Attention is given to uncertainty associated with numerical spatial model inputs and effect of spatial correlations on the results of an uncertainty propagation analysis, but we will demonstrate that the methodology is universal and applies to different data types and models (for example, hydrological, water quality, ecological). Course participants will be provided with R scripts which will allow them to use the methods presented in their research afterwards.