Simulating rice in farming systems - modelling transitions between aerobic and ponded soil environments in APSIM

Adapting farming systems to reduced availability of irrigation water is an emerging research issue in irrigated districts throughout the globe. Water shortages in parts of the rice-growing world have prompted research into a range of alternate agricultural practices, including expansion of rice as a component in diverse farming systems, in rotation with dryland and aerobically irrigated crops and pastures, utilising a range of modified tillage and residue management practices. Evaluation of potential future adaptation strategies can be assisted by well-tested farming systems models that capture interactions between soil water and nutrient dynamics, crop growth, climate and management inputs/practices. APSIM represents such a model, however due to its 'dryland heritage' APSIM has previously been unequipped to describe the soil water, carbon and nitrogen dynamics as soil environments progress from aerobic to anaerobic and back again, such as occurs in crop rotations involving ponded rice and other non-ponded crops (wheat, maize, legumes, pastures etc.). Various relevant chemical and biological processes that occur in long-term ponded water were also unaccounted for in APSIM. In this paper we describe how we incorporated this new functionality into APSIM. This includes evaluation of previous efforts to model anaerobic soil environments; conceptualization of the required new functionality within the existing APSIM framework; and design of code structure to facilitate seamless transitions between aerobic and anaerobic soil environments in continuous simulation. Included is the development of a new module, APSIM-Pond, to describe biological and chemical processes responsible for system loss/gain of C and N in rice ponds. We conclude by providing some preliminary testing results. As a starting point we use many process descriptions from the CERES-Rice and RIWER models. However a new element of APSIM-Pond is our representation of algal turnover and biomass soil incorporation in rice-based farming systems. This is in response to criticism of CERES-Rice's ability to capture long-term trends in soil organic carbon. In simulation of long-term (35 yrs +) ponded rice experiments at IRRI, CERES-Rice simulated a rundown in soil organic carbon, when in fact none was measured. We have included algal contributions to the C & N pools in response to this criticism. No previous modelling framework has addressed the issue of switching between aerobic and anaerobic environments during a simulation, which is particularly important if the focus of the modelling exercise is evaluating new farming system practices that include ponded rice in rotation with non-flooded crops. The ORYZA2000 rice model was previously incorporated into APSIM, and tested in both N- and water-limited environments. This testing highlighted the needs for the improvements to the C & N balances reported here. Similarly, the simulation of greenhouse gas emissions associated with changed practices in rice-based farming systems also requires sensible accounting for C & N in pond and soil. To date, the model developmental work reported in this paper has focused on the soil C & N story with respect to crop production - a future imperative will be partitioning modeled losses sensibly into the relevant greenhouse gases.