The FvCB model is more complicated than it looks for parameter estimation. It has several characteristics that invalidate the use of conventional optimization approaches . For example, the FvCB model is a change-point model and is structurally overparameterized in individual carboxylation limitation states with respect to A/Ci measurements. Without recognizing these unique features of the FvCB model, conventional approaches may lead to incorrect parameters and produce false parameter relatinships. EDO takes into consideration the unique structure of the FvCB model and is better suited for analysis of A/Ci curves. For more details, please read Gu et al. (2010).

We welcome users to evaluate the EDO approach and strongly encourage users to contribute their leaf-level measurements towards the development of a Global LeafWeb Database; however, users may still use the interface without their data becoming part of the larger LeafWeb database.

To achieve the best possible results and to enhance the utility of your contributions for others, it is imperative those submitting data provide both the requested core leaf-level variables and the complementary information necessary to properly qualify the leaf or leaves. The desired core and companion variables are specified in the sample input file and descriptive "readme" file for each service.

The following links provide information on preparing these input files:

LeafWeb employs the Exhaustive Dual Optimization (EDO) approach to fitting A/Ci curves for parameters in the Farquhar-von Caemmerer-Berry (FvCB) model (Gu et al. 2010). For each A/Ci curve, four sets of parameters are fitted because a given A/Ci curve might not contain sufficient constraint power for all parameters in the FvCB model.

More detailed explanation of the output files are given below:

- Readme_cntrlbestparameters - This file contains parameters estimated for all A/Ci curves that are submitted in the same batch.
- Readme_cntrlcomparison - This file contains outputs from PISCAL that facilitates examination of how well the fitting was performed.
- Readme_cntrlparameters - This file is an extended version of the output file 'cntrlbestparameters.csv'

- Relatively densely distributed points in the curvature region of an A/Ci curve are crucial for reliable parameter estimation.
- Use data from literature or exploratory measurements to establish the broad shape of A/Ci curves of similarly positioned leaves of the interested species.
- Identify the Ci range within which the curvature region is located.
- Determine the corresponding cuvette CO2 range using either prior knowledge or preliminary measurements. As an initial try, set the cuvette CO2 to be 1.5 - 2 times the desired Ci values.
- Sample the curvature region preferentially.