Cellular Automata (CA) is capable of (re)producing complex spatial-temporal dynamic patterns by using a set of local evolution rules involving the states of immediate neighbouring cells. Classical Cellular Automata consist of a regular lattice of cells, which lead to a geometric staircase representation at the boundaries. In this research, the Unstructured Cellular Automata (UCA) were adopted. Instead of uniform and structured meshes, UCA are based on the unstructured meshes. The Unstructured Cellular Automata (UCA) have the advantages of presenting the complicated geometry and boundaries flexibly, compared with the classical cellular automata, and it can be coupled directly with unstructured hydrodynamic modelling without mapping transformation. The rules of Cellular Automata are local defined, while the physical transport equations are global depended. It is more convenient to using UCA for solving some non-uniform problems (e.g. spiked pollution diffusion). In this paper, the spiked diffusion pattern was captured as an application. The results show that UCA is a suitable scheme for irregular spatial problems.