A system of stochastic differential equations is studied describing a compartmental carbon transfer model that includes uncertainties arising in the model from environmental and photosynthetic effects as well as initial conditions. Justification is given for the modeling of observed times series as Weiner processes. The solution of the resulting system is obtained as a stochastic process, a formulation is given appropriate for obtaining continuity and differentiability with respect to model transfer coefficients, and numerical approximation results are given. Estimation results of coefficients from NEE data are obtained using quasi-Monte Carlo techniques. The error resulting in NEE stochastic models is observed to be approximately Gaussian. This result is used to construct a joint probability density defined on a sample space of transfer coefficients. Finally, the joint probability density function is used with quasi-Monte Carlo to obtain information on transfer coefficients and predicted carbon pools. This information is compared with a priori results obtained without the benefit of NEE data.
- Output-least-squares estimation
- Probabilistic inversion
- Stochastic differential equations
ASJC Scopus subject areas
- Computational Mathematics
- Applied Mathematics