|Pachepsky, Yakov - DUKE UNIVERSITY|
Submitted to: Ecological Modeling
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: February 1, 1996
Publication Date: N/A
Interpretive Summary: Crop models have become useful tools to evaluate and compare management practices with respect to the impact on crop yield and environmental quality. One drawback of crop models is they often contain soils models where the level of detail is very low. As a result, simulation results with respect to environmental quality may not be satisfactory. Another drawback is that existing crop simulation models are very rigid, which makes it difficult to add new management practices and difficult to add new knowledge as it becomes available. We have developed a generic, modular, two-dimensional soil simulator that can be easily interfaced with plant models. The structure of the model is designed so the code for any particular process can be easily replaced by new code without requiring any restructuring of the model. Code to simulate management practices or the addition of new processes can be easily added as new modules. The model includes water, chemical, gas and heat transport modules, root growth and water uptake modules, and chemical and nitrogen transformation modules. Documentation and source code (FORTRAN) as well as examples are available from the authors.
Technical Abstract: The ability to build models to simulate a range of crop management scenarios can be enhanced by using reputable modules corresponding to soil, root atmosphere and management processes. In this paper, we describe the design of a generic modular soil and root process simulator for use in crop modeling. A three-level hierarchy has been found sufficient to map the knowledge of soil processes and their parameters onto a modularized structure. Decoupling of processes and rearrangement of boundary condition formulations are used to facilitate independence of modules. Encapsulation of information is consistently implemented. Variable time steps are selected using restrictions imposed by non-linearity of models. The representative 2DSOIL simulator is written in FORTRAN. The use of this modular design allows modeler to reuse well-tested codes, and results in the decrease of efforts in input data preparation and in expandability to encompass various management practices. Because of the modular design, 2DSOIL can be readily modified and easily incorporated into crop models.