Multiscale Models of Photosynthesis
From BioMASS Laboratory Wiki
Abstract for 2007 ASABE conference in Minnesota
Determination of constants of a two-leaf multi-scale photosynthesis model using a genetic algorithm
Many models have been developed to estimate plant photosynthesis with the most cited, a biochemical based leaf scale model, being introduced by Farquhar in 1980. Since then there has been considerable effort to integrate such lower scale models into canopy scale models to simulate photosynthesis in a complete plant or a collection of plants. One of the more popular approaches to integrate the models is the two-leaf model (Sinclair) in which the canopy is assumed to be composed on only two leaves, one shaded and one un-shaded (sun) leaf. The PAR (Photosynthetically Active Radiation) and photosynthetic capacity are computed for these two leaves by numerical integration and the Farquhar model using these as input calculates the assimilated carbon for the entire canopy.
The problem with this model is that first the numerical integration involves many equations with several constants and second the assimilated carbon obtained is a daily estimate i.e. it simulates the behavior of the plant for a day and not over the entire life of the plant. We have proposed a model similar to the two–leaf model, but with fewer equations and constants that predicts the daily assimilated carbon for all days from emergence to senescence. The data that we require the model to fit was obtained from growth chamber experiments carried out by several research groups. Since the model has been modified the constants will have to be re-calculated, but the non-linear form of the equations negates the use of linear regression methods.
Genetic algorithms are heuristic search techniques which can be used for regression analysis, by searching for the set of constants that minimize the deviation of the model from the experimental data. Here we are investigating the use of a genetic algorithm to determine the constants of the modified two-leaf model using the original Farquhar model as underlying bio-molecular model.
