Setting parameters for the intamap package

Share:

Description

This function sets a range of the parameters for the intamap package, to be included in the object described in rtop-package

Usage

1
getRtopParams(params,newPar, observations, formulaString, ...)

Arguments

params

An existing set of parameters for the interpolation process, of class
intamapParams or a list of parameters for modification of the default parameters

newPar

A list of parameters for updating params or for modification of the default parameters. Possible parameters with their defaults are given below

observations

SpatialPolygonsDataFrame with observations, used for setting some of the default parameters

formulaString

formula that defines the dependent variable as a linear model of independent variables, see e.g. createRtopObject for more details.

...

Individual parameters for updating params or for modification of the default parameters. Possible parameters with their defaults are given below

  • model = "Ex1" - variogram model type. Currently the following models are implemented:

    • Exp - Exponential model

    • Ex1 - Multiplication of a modified exponential and fractal model, the same model as used in Skoien et al(2006).

    • Gau - Gaussian model

    • Ga1 - Multiplication of gaussian and fractal model

    • Sph - Spherical model

    • Sp1 - Multiplication of spherical and fractal model

    • Fra - Fractal model

  • parInit - the initial parameters and the limits of the variogram model to be fitted, given as a matrix with three columns, where the first column is the lower limit, the second column is the upper limit and the third column are starting values.

  • nugget = TRUE - logical; if TRUE nugget effect should be estimated

  • unc = TRUE - logical; if TRUE the standard deviation of observations are in column unc

  • rresol = 100 - minimum number of discretization points in each area

  • hresol = 5 - number of discretization points in one direction for elements in binned variograms

  • cloud = FALSE - logical; if TRUE use the cloud variogram for variogram fitting

  • amul = 1 - defines the number of areal bins within one order of magnitude. Numbers between 1 and 3 are possible, as this parameter refers to the axp parameter of axTicks.

  • dmul = 3 - defines the number of distance bins within one order of magnitude. Numbers between 1 and 3 are possible, as this parameter refers to the axp parameter of axTicks.

  • fit.method = 8 - defines the type of Least Square method for fitting of variogram. The methods 1-7 correspond to the similar methods in gstat.

    • 1 - weighted least squares with number of pairs per bin:
      err = n * (yobs-ymod)^2

    • 2 - weighted least squares difference according to Cressie (1985):
      err2 = abs(yobs/ymod-1)

    • 6 - ordinary least squares difference: err = (yobs-ymod)^2

    • 7 - similar to default of gstat, where higher weights are given to shorter distances err = n/h^2 * (yobs-mod)^2

    • 8 - Opposite of weighted least squares difference according to Cressie (1985): err3=abs(ymod/yobs-1)

    • 9 - neutral WLS-method - err = min(err2,err3)

  • gDistEst = FALSE - use geostatistical distance when fitting variograms

  • gDistPred = FALSE - use geostatistical distance for semivariogram matrices

  • gDist - parameter to set jointly gDistEst = gDistPred = gDist

  • nmax = 10for local kriging: the number of nearest observations that should be used for a kriging prediction or simulation, where nearest is defined in terms of the space of the spatial locations. By default, 10 observations are used.

  • maxdist = Inf - for local kriging: only observations within a distance of maxdist from the prediction location are used for prediction or simulation; if combined with nmax, both criteria apply

  • hstype = "regular" - sampling type for binned variograms

  • rstype = "rtop" - sampling type for the elements, see also rtopDisc

  • nclus = 1- number of CPUs to use if parallel processing is wanted; nclus = 1 means no parallelization

  • cnAreas = 100- limit whether parallel processing should be applied; the minimum number of areas in varMat, and also controlling when to use parallel processing in rtopDisc, when
    nAreas*params$rresol/100 > cnAreas

  • clusType = NULL- the cluster type to be started for parallel processing; uses the default type of the system when clusType = NULL

  • outfile = NULLfile where output can be printed during parallel execution

  • varClean = FALSElogical; if TRUE it will remove highly correlated areas from the covariance matrix during simulation

  • wlim = 1.5 - an upper limit for the norm of the weights in kriging, see rtopKrige

  • wlimMethod = "all"which method to use for reducing the norm of the weights if necessary. Either "all", which modifies all weights equally or "neg" which reduces negative weights and large weights more than the smallest weights

  • cv = FALSE - logical; for cross-validation of observations

  • debug.level = 1 - used in some functions for giving additional output. See individual functions for more information.

  • partialOverlap = FALSEwhether to work with partially overlapping areas

  • olim = 1e-4smallest overlapping area to be used for partial overlap, relative to the smallest of the areas

  • nclus = 1option to use parallel processing, nclus > 1 defines the number of workers to be started

  • clusType = NAwhich cluster type to start if nclus > 1; the default is used if nclusType = NA

  • cnAreas = 200The minimum number of observations or observations plus predictions allowing parallelization in the creation of the covariance matrix

  • cDlim = 1e6The minimum number of discretization points for allowing parallelization in the discretization process

  • observations - used for initial values of parameters if supplied

  • formulaString - used for initial values of parameters if supplied

Value

A list of the parameters with class rtopParams to be included in the object described in rtop-package

Note

This function will mainly be called by createRtopObject, but can also be called by the user to create a parameter set or update an existing parameter set. If none of the arguments is a list of class rtopParams, the function will assume that the argument(s) are modifications to the default set of parameters. The function can also be called by other functions in the rtop-package if the users chooses not to work with an object of class rtop.

If the function is called with two lists of parameters (but the first one is not of class rtopParams) they are both seen as modifications to the default parameter set. If they share some parameters, the parameter values from the second list will be applied.

Parallel processing has been included for some of the functions. The default is no parallel procesing, and the package also attempts to decide whether it is sensible to start a set of clusters and distribute jobs to them based on the size of the job. The default limit might not be the best for every system.

Author(s)

Jon Olav Skoien

References

Cressie, N. 1985. Fitting variogram models by weighted least squares. Mathematical Geology, 17 (5), 563-586

Skoien J. O., R. Merz, and G. Bloschl. Top-kriging - geostatistics on stream networks. Hydrology and Earth System Sciences, 10:277-287, 2006

See Also

createRtopObject and rtop-package

Examples

1
2
3
4
5
6
7
# Create a new set of intamapParameters, with default parameters:
params = getRtopParams()
# Make modifications to the default list of parameters
params = getRtopParams(newPar = list(gDist = TRUE, nugget = FALSE))
# Make modifications to an existing list of parameters
params = getRtopParams(params = params, newPar = list(gDist = TRUE,
         nugget = FALSE))