Submitted to: Fractals In Soil Science
Publication Type: Proceedings
Publication Acceptance Date: 12/1/2003
Publication Date: 4/4/2004
Citation: Pachepsky, L., Kaul, M., Walthall, C., Daughtry, C., Lyndon, J. 2004. Need and feasibility of applying L-System models in agricultural crop modeling. In: Proceedings of Fractional 2004 Conference, April 3-8, 2004, Vancouver, Canada. p. 317-318. Interpretive Summary: Crop simulation models are useful tools for crop management and yield prediction. Output from the models are typically tables of numbers without graphical displays that can be difficult to interpret. Validating the models involves measurements of actual plants that often do not adequately characterize morphology. An L-systems model producing a graphical depiction of simulated plants would provide a simple reference for comparison with photographs or field observations for validation, and may enhance model use and model output interpretation. An L-systems visual model was developed for soybean from data of two cultivars grown under controlled conditions. Periodic morphological observations and photographs were used to create virtual plants. A simulation model of soybean vegetative development was calibrated for these two cultivars and linked with the visual model. The virtual plants visually compared well with the series of photographs of the real plants. This suggests that L-systems models can play a useful role in visually validating the phenology modules of crop models, and may provide a more meaningful user interface. Despite its apparent usefulness, an obstacle for using L-systems with crop simulation is the absence of a manual suitable for use by a wide range of researchers working with crop models.
Technical Abstract: Contemporary agriculture uses crop simulation models for crop management and yield prediction. However, model validation remains a problem, and user interfaces contain only quantitative information. An L-systems model coupled with a crop model could provide an additional means of visual validation and serve as an attractive, interactive, and practical user interface. Detailed morphological observations and photographs were taken from emergence to seed filling and were used to create virtual plants. A functional model of vegetative development [1, 2] was parameterized for these two cultivars and linked with the visual model. The results showed that the virtual plants visually compared well with the series of photographs of the real plants. This suggests that L-systems models can play a useful role in visually validating the phenology modules of crop models, and can provide a meaningful user interface. The software used for the visual modeling is an extraordinarily powerful tool. However, a serious obstacle for using L-systems for crop simulation is the absence of a manual suitable for use by a wide range of researchers working with crop models.