Padmaja Ramankutty

Auto-calibrating emulation models for predicting perennial pasture production

APSIM, a software program of a biological model, simulates growth of plants under varying conditions. Initially used for wheat, it has been extended to other annual crops and recently to perennial pasture plants.

The model has numerous inputs and detailed outputs. APSIM can be used to determine the optimal environmental conditions under which a particular plant will give maximum growth. This requires exhaustive simulations of plant growth under each of the environments, which entails summarizing huge datasets.

Padmaja’s research is creating a statistical model which is a simplified version of APSIM that contains fewer inputs, fewer outputs, and with less complex relationships than those in APSIM. This model will still produce total production predictions that are not significantly different statistically from those generated by APSIM.

Padmaja’s research falls into the CRC’s Program 2 research area and is related to the EverCrop and EverDecide project.

Objectives:

• Examine the relationships between factors that influence plant growth, and APSIM-generated biomass production
• Propose parsimonious statistical models to predict APSIM-generated biomass production
• Develop methods to auto-calibrate these statistical models using historical site-specific or species-specific data
• Assess capability of final models to predict biomass production of real farm or experimental data for existing and newly developed perennial pasture varieties.

For more information, email Padmaja