Modelling water release and absorption in soils using cellular automata

Abstract

Many important microbial, chemical and physical processes in soils take place within water-filled pores. Whilst the behaviour of chemicals and micro-organisms in aqueous conditions has often been thoroughly researched, their behaviour in soils is still difficult to predict. One approach to this is to superimpose models of the behaviour of chemicals and micro-organisms in water onto a model of the behaviour of water in soil. In this paper we set out to demonstrate the potential to use a simple cellular automaton rule set to describe the location of water in a soil matrix under varying conditions of matric suction for possible future combining with other process models. The model used a three-dimensional array of pixels (which could be solid or void) to represent the structure of a small volume of soil. Parameters describing the porosity and the proportion of void pixels directly adjacent to solid pixels were used to define the structure. These were derived from image analysis of resin impregnated samples of a silty clay, silt loam and sandy loam. Ballotini beads were also used as a further “soil type”. Water retention under a range of increasing suctions (1–300 kPa) was modelled and compared with measurements made using conventional methods on the same soils. The cellular automaton rule set generally produced a good description of water release from soils. Errors were largely attributable to limits on the maximum and minimum particle sizes which could be represented in the present model due to limitations of computing power. This approach to modelling soil water offers many opportunities for understanding complex interactions between soil water and other processes occurring in soils.