No abstract

Biomass and nitrogen partitioning Integrated design environment Phenological and morphological development Reusable organ and function classes a b s t r a c tThe Plant Modelling Framework (PMF) is a software framework for creating models that represent the plant components of farm system models in the agricultural production system simulator (APSIM). It is the next step in the evolution of generic crop templates for APSIM, building on software and science lessons from past versions and capitalising on new software approaches. The PMF contains a top-level Plant class that provides an interface with the APSIM model environment and controls the other classes in the plant model. Other classes include mid-level Organ, Phenology, Structure and Arbitrator classes that represent specific elements or processes of the crop and sub-classes that the mid-level classes use to represent repeated data structures. It also contains low-level Function classes which represent generic mathematical, logical, procedural or reference code and provide values to the processes carried out by mid-level classes. A plant configuration file specifies which mid-level and Function classes are to be included and how they are to be arranged and parameterised to represent a particular crop model. The PMF has an integrated design environment to allow plant models to be created visually. The aims of the PMF are to maximise code reuse and allow flexibility in the structure of models. Four examples are included to demonstrate the flexibility of application of the PMF; 1. Slurp, a simple model of the water use of a static crop, 2. Oat, an annual grain crop model with detailed growth, development and resource use processes, 3. Lucerne, perennial forage model with detailed growth, development and resource use processes, 4. Wheat, another detailed annual crop model constructed using an alternative set of organ and process classes. These examples show the PMF can be used to develop models of different complexities and allows flexibility in the approach for implementing crop physiology concepts into model set up.

Soil water is the single most important resource for pasture and crop production in New Zealand farms. Because soil water is difficult to measure, however, the ability to predict soil water status from daily weather data is valuable, and has application for on-farm irrigation, stocking, and supplementation decisions. In this paper a practical water balance model is presented. The model uses daily rainfall and potential evapotranspiration (PET) estimates to predict changes in the water content in two overlapping soil zones: a rapidly recharged (and depleted) zone of unspecified depth, and the total plant rooting zone. The use of two zones improves predictions of actual evapotranspiration and plant stress compared with models that use only one zone. An important factor determining the success of soil water models is the ability to predict actual evapotranspiration, AET. In this model actual evapotranspiration, AET, is calculated as the lesser A00039 Received 4 August 2000; accepted II December 2000of potential evapotranspiration, PET, and total readily available water (RAW) per day. RAW is defined as all of the water in the rapidly recharged surface zone plus a proportion of the water in the remainder of the soil profile. By validation against 11 historical data sets, the model is shown to give accurate predictions of soil water deficit across a range of New Zealand flat-land pastoral soils. The model parameters can be easily estimated from commonly available soil properties (soil order classification, and available water holding capacity) without the need for additional site-specific calibration. This model provides an easily used, practical decision tool for the management of drought, allowing early prediction of decline in pasture growth and estimates of required irrigation.

Fertiliser management is an important aspect of growing good forage brassica crops. Every crop has a different requirement, depending on soil fertility and the expected yield response. Systems were developed for forecasting how much fertiliser, and what types, to apply to individual kale and Pasja crops. First, yield responses to fertiliser application were measured in trials in diverse climates and soil fertility conditions. Yield responded strongly to N and P availability, there were few responses to K fertiliser application, and there were no responses to S application. Second, overall responses to the nutrient supply from soil and fertiliser sources were determined in a comprehensive across-trials analysis using the PARJIB model. R-squared values for correlations between actual yields and yields simulated with the PARJIB calibrations were 0.65 and 0.64 for Pasja and kale respectively. Finally, the results were programmed into new software systems (The Kale Calculator and The Pasja Calculator) that deliver a forecast for each crop of the types and amounts of fertiliser that will give the best economic return on the investment in fertiliser. Keywords: The Kale Calculator, The Pasja Calculator, fertiliser application, yield response, PARJIB analysis

No abstract