pfDotMap: Produce maps of paleofire data

In paleofire/paleofire: Analysis of Charcoal Records from the Global Charcoal Database

Description

Produce map graphics representing spatial variability in charcoal data from the Global Charcoal Database.

Usage

pfDotMap(
  TR,
  tarAge,
  hw,
  binhw = 0.5 * mean(diff(tarAge)),
  fig.base.name = NULL,
  base.map = "coasts",
  grd.res = 5,
  grd.ext = c(-180, 180, -90, 90),
  grd.lonlat = NULL,
  proj4 = "+proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs",
  n.boot = 1000,
  cx.minsize = 0.3,
  cx.mult = 1
)

Arguments

TR	An object returned by pfTransform
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).
hw	Numeric, the half window width for the locfit procedure (in years).
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).
fig.base.name	Character sequence representing the base name for the figures. Can be preceded by a path as long as all directories in the path exist. One figure will be produced for each time bin, with years (and file suffix) appended to the base name automatically. A value of NULL (default) causes figures to be plotted to the current device in sequence.
base.map	Currently, either 'coasts' or 'countries' to choose which base map (from required library 'rworldmap') to be plotted as the base map for all plots. Could easily be modified to accept any SpatialPolygons object.
grd.res, grd.ext	Desired grid resolution and extent in degrees. If grd.res is a single number, the grid will be defined with equal lon/lat resolution; a two-element vector (lon,lat) can also be supplied for unequal resolution. grd.ext is specified as a vector of the form c(min-lon,max-lon,min-lat,max-lat).
grd.lonlat	A data frame of coordinates for every grid cell center, to be used in cases where an irregular grid is desired. Columns must be named 'lon' and 'lat'. If specified, grd.res and grd.ext are ignored. Note that this option could have undesirable results for unusual grid definitions. In particular, the maximum radius for including sites in a grid cell is always calculated at the equator. For a regular lon/lat grid, this guarantees all sites will be included in at least one cell, because equatorial cells are largest at the equator. If an irregular grid is specified such that this is not true, the maximum radius calculated could lead to sites excluded from all cells. In this case a warning is printed but the function proceeds anyway.
proj4	proj.4 string representing the desired projection for plotted maps. Default is unprojected. See http://www.spatialreference.org to look up the string for your favorite projections.
n.boot	Number of bootstrap replicates to use when creating confidence intervals around each grid-cell mean. In each time bin X grid cell combination, replicates consist of composite z-score values for that bin, randomly sampled (with replacement) from sites within the grid cell (see 'Details' for precise description of sites included in each cell). I.e., no temporal bootstrapping is done here, so that bootstrap CI reflect only spatial variability.
cx.minsize, cx.mult	Parameters that crudely adjust plotted dot size. cx.minsize defines the minimum cex applied to any point in any map, cx.mult scales all points by an equivalent factor.

Details

Takes any pfTransform object as input, and allows any set of one or more time bins to be specified for plotting (one plot per bin). Time bins are specified as for pfCompositeLF (which is called by pfSimpleGrid. The extent, resolution, and projection of the desired grid are also user-specified.

Results will be plotted on a regular lon/lat grid. To determine which sites contribute to each grid cell value, the code searches within a specified great circle distance (i.e. on the surface of the globe) around each grid cell center. To avoid missing any sites, the distance is set equal to the greatest distance from a grid cell center to its most distant corner, which occurs at the equator where grid cells are largest. This conservative approach will result in many sites falling within multiple grid boxes. At all latitudes, the defined radii will overlap near the edges of the grid boxes. At higher latitudes, the lon/lat grid cells are physically much smaller, so overlap will be considerably greater. There are alternatives, like using a grid that is irregular in terms of lon/laton, or changing the area of grid cells depending on latitude. But all have their tradeoffs, and this one is simple.

Current version produces plots of mean CHAR, number of sites per grid cell, and number of grid cells contributed to by each site (due to overlapping radii described above). The mean plot additionally shows points in two sizes, representing those mean values whose 95"%" confidence intervals do (small dots) or do not (large dots) contain zero. Finally, a time series is plotted in each figure with the current time bin highlighted.

Value

Plots are produced on the current device or in pdf files defined by fig.base.name. In addition, a named list of useful objects is returned:

COMP	The binned composite generated for plotting.
bins	The list of bin endpoints.
sp.grd	A SpatialPointsDataFrame-class object containing all the grid-level statistics produced and plotted (mean influx value, bootstrap confidence interval, and number of sites per grid cell).
sp.sites	A SpatialPointsDataFrame-class object representing the number of grid cells influenced by each site.
plots	A list with one element for each bin. These elements are themselves named lists of trellis objects representing each of the plots produced ("mean", "sitesPerCell", "cellsPerSite", "timeSeries"). Note that these objects can be edited to some degree with the update.trellis function, and plotted or used in layouts as any other trellis graphics can.

Author(s)

R. Kelly

References

Power, M., J. Marlon, N. Ortiz, P. Bartlein, S. Harrison, F. Mayle, A. Ballouche, R. Bradshaw, C. Carcaillet, C. Cordova, S. Mooney, P. Moreno, I. Prentice, K. Thonicke, W. Tinner, C. Whitlock, Y. Zhang, Y. Zhao, A. Ali, R. Anderson, R. Beer, H. Behling, C. Briles, K. Brown, A. Brunelle, M. Bush, P. Camill, G. Chu, J. Clark, D. Colombaroli, S. Connor, A. L. Daniau, M. Daniels, J. Dodson, E. Doughty, M. Edwards, W. Finsinger, D. Foster, J. Frechette, M. J. Gaillard, D. Gavin, E. Gobet, S. Haberle, D. Hallett, P. Higuera, G. Hope, S. Horn, J. Inoue, P. Kaltenrieder, L. Kennedy, Z. Kong, C. Larsen, C. Long, J. Lynch, E. Lynch, M. McGlone, S. Meeks, S. Mensing, G. Meyer, T. Minckley, J. Mohr, D. Nelson, J. New, R. Newnham, R. Noti, W. Oswald, J. Pierce, P. Richard, C. Rowe, M. Sanchez Goni, B. Shuman, H. Takahara, J. Toney, C. Turney, D. Urrego-Sanchez, C. Umbanhowar, M. Vandergoes, B. Vanniere, E. Vescovi, M. Walsh, X. Wang, N. Williams, J. Wilmshurst, and J. Zhang. 2008. Changes in fire regimes since the Last Glacial Maximum: an assessment based on a global synthesis and analysis of charcoal data. Climate Dynamics 30:887-907.

Examples

## Not run: 
## Composite charcoal record for North America:
ID=pfSiteSel(continent=="North America", long<(-100),l12==1 & long<(-130))
plot(ID)

## Transform data
res3=pfTransform(ID,method=c("MinMax","Box-Cox","Z-Score"),BasePeriod=c(200,4000))

## Plot maps for 1000-yr bins spanning 3-0 kBP
# dev.new(width=10,height=10) # A big plot area helps. 
dotmap = pfDotMap( TR=res3, tarAge=seq(0,2000,1000), hw=500, grd.ext=c(-170,-80,40,80), 
                   cx.minsize=2,cx.mult=3)
summary(dotmap)

# Plot the mean map from the first time bin
# newmap = update(dotmap$plots[[1]]$mean, main="A relabeled map")
# newmap

## End(Not run)

paleofire/paleofire documentation built on Dec. 29, 2021, 11:09 a.m.