Spatio-Temporal Simulation and Sensitivity Analysis of Oil Spill CA Model

In order to overcome the difficulty of obtaining parameters incellular automata (CA) model for oil spill,this paper introduces logistic regression method into CA model and presents a new oil spill model based on logistic-regression CA model.This model can easily obtain model parameters and simulate the dynamic changes of oil spill by using only a few inputs, such as the initial image, impact factor.The model was applied to DeepSpill experiment to verify its effect in simulation of oil spill. Experiments show that the simulation results are consistent with the real situation.The total accuracy and Kappa coefficient of simulation results is 96.6% and 0.899 respectively. After the sensitivity analysis of parameters, we found that the impact of temperature and salinity on simulation results is the weakest, followed by proximity variable. The most important variables in the model are the winds and currents. If we don't take winds and currents into consideration, the simulation results will deviate from the verification image for lacking the important characteristics-drift behavior of oil spill.