agelme: Estimation of the relationship between Calibrated age and...

Description Usage Arguments Details Value Author(s) References Examples

Description

Estimates the relationship of Calibrated age and depth for paleorecords. The function uses a smooth spline of the mgcv library by Simon Wood. It produces predicted confidence interval for the relationship approximating a mixed effect model, as there are two levels of uncertainty, i.e. within dated object and between dated objects.

Usage

1
2
agelme(depup, depdo, bpup, bpdo, use, weights=c(1,rep(0,length(depup)-1)), 
    vspan=1, k=length(depup)-1, m=2, diagnostic=FALSE)

Arguments

depup

The upper depths of the dated slides

depdo

The lower depths of the dated slides

bpup

The younger calibrated ages of the dated slides

bpdo

The older calibrated ages of the dated slides

use

Logical vector of dates to include in the model. Default is to use all.

weights

Weights to be used for the estimation, default is fixed top-layer followed by inverse variance of within dated object

vspan

The span to be used for the diagnostic plots, default span = 1

k

Number of base function to start the shrinkage in the gam estimation procedure

m

The order of penalty for the term, i.e. the degree of continuity at the knots (default, m = 2 gives cubic smooth spline)

diagnostic

Logical, should diagnostic plots be made.

Details

Note that the fixation of the top layer is done by a weight = 1, whereas the other weights follows inverse variance within object.

The diagnostic plots is used to check the quality of the estimation and to see if there is a need for an assumption of between object variance proportional to mean. The latter however is rarely encountered for paleodata.

Value

tdf

Degrees of freedom used by the cubic smooth spline, a vector with first value for constant variance and second vector for variance equal to mu.

weights

A vector of the weights used by the cubic smooth spline

RES

A vector of the Residual sum of squares

Models

A list with the models from the cubic smooth spline, constant and mu variance, respectively

Data

A data.frame including the data used for the estimation

Author(s)

Einar Heegaard <[email protected]>

References

Heegaard, E., Birks, HJB. & Telford, RJ. 2005. Relationships between calibrated ages and depth in stratigraphical sequences: an estimation procedure by mixed-effect regression. The Holocene 15: 612-618

Examples

1
2
3
4
5
6
7
8
data(STOR)

fit.mod <- with(STOR,agelme(depthup,depthdo,cageup,cagedo))

#Predicting using the constant variance model,
#for each cm between 70 and 400 cm.
fit.pre <- predict(fit.mod,1,70:400)
plot(fit.pre)

palaeoSig documentation built on May 29, 2017, 9:11 a.m.