Markov chain Monte Carlo for C4 photosynthesis parameters

Usage

MCMCc4photo(data, niter = 20000, ivmax = 39, ialpha = 0.04, ikparm = 0.7, itheta = 0.83, ibeta = 0.93, iRd = 0.8, Catm = 380, b0 = 0.08, b1 = 3, StomWS = 1, ws = c("gs", "vmax"), scale = 1, sds = c(1, 0.005), prior = c(39, 10, 0.04, 0.02), UPPERTEMP = 37.5, LOWERTEMP = 3)

Arguments

data
observed assimilation data, which should be a data frame or matrix. The first column should be observed net assimilation rate (micro mol per meter squared per secondmicro mol per meter squared per second). The second column should be the observed quantum flux (micro mol per meter squared per secondmicro mol per meter squared per second). The third column should be observed temperature of the leaf (Celsius). The fourth column should be the observed relative humidity in proportion (e.g. 0.7).
niter
number of iterations to run the chain for (default = 20000).
ivmax
initial value for Vcmax (default = 39).
ialpha
initial value for alpha (default = 0.04).
ikparm
initial value for kparm (default = 0.7). Not optimized at the moment.
itheta
initial value for theta (default = 0.83). Not optimized at the moment.
ibeta
initial value for beta (default = 0.93). Not optimized at the moment.
iRd
initial value for dark respiration (default = 0.8).
Catm
see c4photo function.
b0
see c4photo function.
b1
see c4photo function.
StomWS
see c4photo function.
ws
see c4photo function.
scale
This scale parameter controls the size of the standard deviations which generate the moves in the chain.
sds
Finer control for the standard deviations of the prior normals. The first element is for vmax and the second for alpha.
prior
Vector of length 4 with first element prior mean for vmax, second element prior standard deviation for vmax, third element prior mean for alpha and fourth element prior standard deviation for alpha.

Value

an object of class MCMCc4photo with components

Description

This function implement Markov chain Monte Carlo methods for the C4 photosynthesis model of Collatz et al. The chain is constructed using a Gaussian random walk. This is definitely a beta version of this function and more testing and improvements are needed. The value of this function is in the possibility of examining the empirical posterior distribution (i.e. vectors) of the vmax and alpha parameters. Notice that you will get different results each time you run it.

References

Brooks, Stephen. (1998). Markov chain Monte Carlo and its application. The Statistician. 47, Part 1, pp. 69-100.

Examples

## <strong>Not run</strong>: # ## Using Beale, Bint and Long (1996) # data(obsBea) # # ## Different starting values # resB1 <- MCMCc4photo(obsBea, 100000, scale=1.5) # resB2 <- MCMCc4photo(obsBea, 100000, ivmax=25, ialpha=0.1, scale=1.5) # resB3 <- MCMCc4photo(obsBea, 100000, ivmax=45, ialpha=0.02, scale=1.5) # # ## Use the plot function to examine results # plot(resB1,resB2,resB3) # plot(resB1,resB2,resB3,plot.kind='density',burnin=1e4) # # ## <strong>End(Not run)</strong>