The motivation for this work is to create a simple diagnostic model of gross primary production (GPP). The purpose of this is meant to contrast the complex process-based models that are currently being developed and used in earth system modelling that are generally not well tested and tend to produce widely divergent results.
Our data-based model is founded on in situ measurements from the free and fair use FLUXNET archives (http://fluxnet.ornl.gov/obtain-data). We drive our model using basic geophysical and meteorological observations provided by a variety of remote sensing products. Lastly, our model is founded on the principle of optimal acclimation of photosynthesis to the environment.
The key outputs of our modeling work include the daily, monthly and annual GPP; the monthly light-use efficiency (LUE); a generalized LUE model; and, based on the LUE model, the ability to extrapolate GPP globally with the possibility of doing some predictions of future climate scenarios.
The model is based on an open-source framework that consists of a database that stores all the observations and measurements to provide unified data access and a Python-based application code, which is written to be readable and reproducible while still being computationally powerful.
The challenge is to create a general LUE model. Starting with the basic definition of light-use efficiency, the ratio of monthly GPP to monthly absorbed light, this can be expressed as a linear equation where monthly GPP is equal to the light-use efficiency multiplied by the monthly absorbed light. This ‘basic’ model, for some flux towers, is capable of explaining 90% or more of variance in GPP; however, it is apparent that for other flux towers something is still missing.
This ‘missing information’ helped motivate the creation of an advanced LUE model, based on the optimality principle with consideration to temperature, elevation, and water-stress effects on GPP. This new theoretically-based observationally-constrained model for terrestrial GPP has promising implications for the future of vegetation and climate modeling.