An R package for generating predictions of whole tree sap flow based on sap flux density observations.
Installing the function for testing in R is easy:
# install.packages("devtools")
devtools::install_github("berdaniera/sapflux")Usage is equally easy. Once installed, for an example, try:
library(sapflux)
# Example raw data
# sap flux density observations (g/m^2/s) from 0-2 cm in xylem
# for a ring-porous tree over 13 days.
plot(egtree,type="b",ylab="Sap flux density (g/m^2/s")
egpred <- qtot(egtree, 0, 0.02, "Ring-porous", FALSE, 1, 0.255, NULL)
plot(egpred,type="l",ylab="Sap flow (g/s)")The function for generating an estimate of whole-tree water use rates is qtot:
qtot(v, a=0, b, woodType=c("Tracheid","Diffuse-porous","Ring-porous"), uncertainty=FALSE, nboot=5000, treeRadius, sapRadius=NULL)Here v is the observed sap flux density observation, a and b are the start and end depths of the sap flux probe (in m), woodType is the xylem anatomy, uncertainty specifies whether to generate uncertainty estimates, nboot is the number of bootstrapped estimates for the uncertainty estimates, treeRadius is the radius of the tree (in m) and sapRadius is an optional depth of the sapwood (in m).
Predictions with uncertainty are based on parameter uncertainty from the model fit to observations. In the 'qtot' function, assign 'uncertainty=TRUE'. This option takes samples from the multivariate posterior distribution for the shape parameters of the radial profile function (instead of using the posterior median). The resulting matrix has a column for each sample, which can be aggregated for calculating intervals:
egpred_uncertainty <- qtot(egtree, 0, 0.02, "Ring-porous", TRUE, 1000, 0.255, NULL)
egpred_interval <- apply(egpred_uncertainty, 1, quantile, c(0.025,0.5,0.975))
plot(egpred_interval[3,], type="l", col="darkgrey", ylab="Sap flow (g/s)")
lines(egpred_interval[1,], col="darkgrey")
lines(egpred_interval[2,])- Author: Aaron Berdanier ([email protected])