optimal_parameters: Model optimization routine

In oscarperezpriego/ETpartitioning: Partitioning of eddy covariance ET into flux components

Description

Routine to estimate optimal parameters of a photosynthesis model using a multi-constraint Markov Chain Monte Carlo (MCMC) (see Perez-Priego et al., 2018).

Usage

optimal_parameters(par_lower, par_upper, data, ColPhotos, ColPhotos_unc,
  ColH, ColVPD, ColTair, ColPair, ColQ, ColCa, ColUstar, ColWS, ColSW_in,
  Chi_o, WUE_o)

Arguments

par_lower	A vector containing the lower bound of the parameters (a1,Do,To,beta)
par_upper	A vector containing the upper bound of the parameters (a1,Do,To,beta)
data	Data.frame or matrix containing all required variables.
ColPhotos	Column name of numeric vector containing time series of photosynthesis data (umol CO2 m-2 s-1).
ColPhotos_unc	Column name of numeric vector containing time series of photosynthesis uncertainties (umol CO2 m-2 s-1).
ColH	Column name of numeric vector containing time series of sensible heat flux (W m-2).
ColVPD	Column name of numeric vector containing time series of vapor pressure deficit (hPa).
ColTair	Column name of numeric vector containing time series of air temperature (deg C).
ColPair	Column name of numeric vector containing time series of atmospheric pressure (kPa).
ColQ	Column name of numeric vector containing time series of photosynthetic active radiation (umol m-2 s-1).
ColCa	Column name of numeric vector containing time series of atmospheric CO2 concentration (umol Co2 mol air-1).
ColUstar	Column name of numeric vector containing time series of wind friction velocity (m s-1).
ColWS	Column name of numeric vector containing time series of wind velocity (m s-1).
ColSW_in	Column name of numeric vector containing time series of incoming short-wave radiation (W m-2).
Chi_o	Long-term effective chi
WUE_o	Long-term effective WUE

Value

a numeric vector containing 4 optimal parameters (a1,Do,To,beta):

Note

The 4 model parameters (a1, Do, Topt and beta, see Perez-Priego et al., 2018) are estimated using a multi-constraint Markov Chain Monte Carlo (MCMC).The objective function (OF) is to find those numerical solutions that minimize not only the mismatch between observed and modeled Photos but also the unit cost of transpiration by introducing a conditional factor demand (phi), which invokes the optimality hypothesis. The phi term is to be defined as the integrated cost of transpiration (i.e. transpiration_mod/photos_mod) over a time period (5 days) normalized by a factor describing the long-term effective water use efficiency (WUE_o).

References

Perez-Priego, O., G. Katul, M. Reichstein et al. Partitioning eddy covariance water flux components using physiological and micrometeorological approaches, Journal of Geophysical Research: Biogeosciences. In press

Examples

 ## Selecting a single day (e.g. 15-05-2011)
 tmp <-  EddySample[ EddySample$TIMESTAMP_START>  201105150000,]
 tmp <-  tmp[tmp$TIMESTAMP_START<  201105160000,]
 ## Defining parameter values

 optimal_parameters(par_lower = c(0, 0, 10, 0)
                    ,par_upper = c(400,0.4, 30, 1)
                   ,data = tmp
                   ,ColPhotos = "GPP_NT_VUT_MEAN"
                   ,ColPhotos_unc = "NEE_VUT_USTAR50_JOINTUNC"
                   ,ColH = "H_F_MDS"
                   ,ColVPD = "VPD_F"
                   ,ColTair = "TA_F"
                   ,ColPair = "PA_F"
                   ,ColQ = "PPFD_IN"
                   ,ColCa = "CO2_F_MDS"
                   ,ColUstar = "USTAR"
                   ,ColWS = "WS_F"
                   ,ColSW_in = "SW_IN_F"
                   ,Chi_o = 0.88
                   ,WUE_o = 24.25)

oscarperezpriego/ETpartitioning documentation built on June 4, 2019, 12:13 a.m.