Steven Jester - April 6

Weather and Climate Systems

Constraint of Vegetation Photosynthesis and Respiration Model (VPRM) parameters using a Markov Chain Monte Carlo (MCMC) technique

Steven Jester

Wednesday, April 6

03:00 PM

Online

The Vegetation and Respiration Model (VPRM) is a land-surface model that simulates the flux of carbon dioxide (CO2). This is done by using certain meteorological parameters such as surface temperature and shortwave radiation as well as vegetation parameters such as water stress and the enhanced vegetation index (EVI) which is a measure of greenness. The VPRM models the net flux of CO2 as the sum of respiration and photosynthesis. The respiration equation as originally developed by Mahadevan et al (2008) was a linear equation. The current model, developed by Gourdji et al (2020) and Hu et al (2021), uses a quadratic fit for temperature and accounts for the water stress and greenness of the vegetation. This was done to better account for nighttime respiration.

Previous studies used a combination of literature values and optimization techniques that incorporated data from eddy covariance towers to optimize the parameters used in the VPRM equations [Hilton et al, 2013; Hu et al, 2021]. This study uses a Bayesian sampling technique, Markov Chain Monte Carlo (MCMC), to perform the parameter optimization. Hilton et al (2013) used a technique similar to MCMC in their optimization, however they performed the optimization on the original linear respiration equation. MCMC is an iterative process that compares a proposed set of parameters against previously accepted parameters to find a set that minimizes a cost function. Preliminary results using this method find that overall fit of parameters over cropland is improved, however the MCMC technique tends to under represent the positive flux. This may mean that this technique is not capturing the diurnal cycle correctly in the optimization.