Latin hypercube sampling and geostatistical modeling of spatial uncertainty in a spatially explicit forest landscape model simulation

Geostatistical stochastic simulation is always combined with Monte Carlo method to quantify the uncertainty in spatial model simulations. However, due to the relatively long running time of spatially explicit forest models as a result of their complexity, it is always infeasible to generate hundreds or thousands of Monte Carlo simulations. Thus, it is of great importance to generate a relatively small set of conditional realizations capturing most of the spatial variability. In this study, we introduced an effective sampling method (Latin hypercube sampling) into a stochastic simulation algorithm (LU decomposition simulation). Latin hypercube sampling is first compared with a common sampling procedure (simple random sampling) in LU decomposition simulation. Then it is applied to the investigation of uncertainty in the simulation results of a spatially explicit forest model, LANDIS. Results showed that Latin hypercube sampling can capture more variability in the sample space than simple random sampling especially when the number of simulations is small. Application results showed that LANDIS simulation results at the landscape level (species percent area and their spatial pattern measured by an aggregation index) is not sensitive to the uncertainty in species age cohort information at the cell level produced by geostatistical stochastic simulation algorithms. This suggests that LANDIS can be used to predict the forest landscape change at broad spatial and temporal scales even if exhaustive species age cohort information at each cell is not available.