ASSESSMENT OF COMPUTATIONALLY INTENSIVE SPATIAL STATISTICAL METHODS FOR GENERATING INPUTS FOR SPATIALLY EXPLICIT ERROR BUDGETS

ABSTRACT. Spatially explicit error budgets are being developed for a process gap-phase forest dynamics model. An error budget is a comprehensive catalog of the different error sources in both surveys and models. When there are errors, it is not intuitive how all these errors propagating through the modeling system will influence final predictions and their uncertainties. The uncertainties due to these errors can also be quite different for small and large areas. Error budgets can help in assessing how various errors propagate through a monitoring/modeling system. They also provide a means for the systematic improvement of a model and its inputs. Practical uses of error-budgets include data correction, model evaluation, quality control, and decision-making with uncertainty, error management, etc. As the first part of developing the spatial error budgets, the spatial inputs for a gap model need to be properly represented in terms of input uncertainty. In this paper, ordinary kriging, sequential Gaussian, and indicator simulation methods were used and compared for spatial prediction of basal area based on a data set from a very intensive survey of stand basal area (400 plots, 20 m by 20 m systematic grid). Stand basal area is one of the inputs needed for the gap-phase model. Both sequential Gaussian and indicato