paleofire (version 1.1.8)

pfCompositeLF: Produce a composite serie from multiple charcoal records using a local regression procedure (from the locfit package)

Description

Produces a composite series from multiple charcoal records by using a robust locally weighted scatterplot smoother (LOWESS). The robust LOWESS uses the locfit function from the locfit package and is applied repeatedly (nboot times) on bootstrapped charcoal sites samples. The records charcoal values are pre-binned prior to sites resampling. This procedure is equivalent to Daniau et al. (2012).

Usage

pfCompositeLF(TR, hw = 250, tarAge = NULL, binhw = NULL, nboot = 1000, conf = c(0.05, 0.95), pseudodata = FALSE, verbose = TRUE)

Arguments

TR

An object returned by pfTransform

hw

Numeric, the half window width for the locfit procedure (in years).

tarAge

Numeric, the target ages for prebinning given in years (e.g. tarAge = seq(0, 10000, 20)). If unspecified the sequence is defined as tarAge=seq(from=min age, to=max Age, by=median resolution).

binhw

Numeric, bin half width for the prebinning procedure (use the same value as tarAge intervals for overlapping bins or tarAge intervals/2 for non-overlapping bins, defautl).

nboot

Numeric, a number specifying the number of bootstrap replicates.

conf

Numeric, define confidence levels.

pseudodata

Logical, if TRUE 10 percent of the data is reflected at the top and the bottom of the resampled serie prior of each locfit regression in order to correct for the edge effect introduced by the local regression, see Cowling & Hall (1996). Equivalent to "minimum slope" correction in Mann(2004).

verbose

Logical: verbose or not...

Value

out

A "pfCompositeLF" object.

References

Daniau, A. L., P. J. Bartlein, S. P. Harrison, I. C. Prentice, S. Brewer, P. Friedlingstein, T. I. Harrison-Prentice, J. Inoue, K. Izumi, J. R. Marlon, S. Mooney, M. J. Power, J. Stevenson, W. Tinner, Andri, M., J. Atanassova, H. Behling, M. Black, O. Blarquez, K. J. Brown, C. Carcaillet, E. A. Colhoun, D. Colombaroli, B. A. S. Davis, D. D'Costa, J. Dodson, L. Dupont, Z. Eshetu, D. G. Gavin, A. Genries, S. Haberle, D. J. Hallett, G. Hope, S. P. Horn, T. G. Kassa, F. Katamura, L. M. Kennedy, P. Kershaw, S. Krivonogov, C. Long, D. Magri, E. Marinova, G. M. McKenzie, P. I. Moreno, P. Moss, F. H. Neumann, E. Norstrom, C. Paitre, D. Rius, N. Roberts, G. S. Robinson, N. Sasaki, L. Scott, H. Takahara, V. Terwilliger, F. Thevenon, R. Turner, V. G. Valsecchi, B. Vanniere, M. Walsh, N. Williams, and Y. Zhang. 2012. Predictability of biomass burning in response to climate changes. Global Biogeochem. Cycles 26:GB4007. Cowling A, Hall P (1996) On pseudodata methods for removing boundary effects in kernel density estimation. Journal of the Royal Statistical Society, Series B 58(3): 551-563. Mann, M. E. (2004). On smoothing potentially non-stationary climate time series. Geophysical Research Letters, 31(7).

Examples

ID=pfSiteSel(id_region=="WNA0", l12==1, long>=-160 & long<=-140)
plot(ID, xlim=c(-180, -130), ylim=c(40,80))
TR=pfTransform(ID, method=c("MinMax","Box-Cox","MinMax","Z-Score"),
               BasePeriod=c(200,2000),QuantType="INFL")

COMP1=pfCompositeLF(TR, tarAge=seq(-50,4000,10), hw=200, nboot=100)

plot(COMP1)

## Note: comparing confidence intervals based on 100 replicates is not recommended
# (100 is used to decrease analysis time)