findLocalBias {intamapInteractive}

Finds (and removes) biases between overlapping networks

Description

The function tries to identify differences between different networks of observation stations that share a region. From these differences, biases are estimated, and can be removed.

Usage

findLocalBias(object, gid = "group",
              formulaString = value ~ 1, regCode="regCode",...)
removeLocalBias(object, localBias, gid = "group", formulaString = value ~ 1, 
                regCode = "regCode")

Arguments

object

data frame with observations

gid

name of column identifying groups of local networks

formulaString

formula that defines the dependent variable as a linear model of independent variables; suppose the dependent variable has name z, for ordinary and simple kriging use the formula z~1; for universal kriging, suppose z is linearly dependent on x and y, use the formula z~x+y

regCode

the column name of regions in the object, if existing

localBias

List of data frames, for a single region, or for each of the regions, each containing biases for different networks in the region(s), result of findLocalBias

...

arguments to be passed to sub-functions

Details

findLocalBias tries to identify biases between overlapping networks, i.e. when there is no boundary between different networks sampling the same type of data. This can typically happen if different governmental bodies are responsible for different types of measurement, e.g. one measuring the situation around populated areas, the other one measuring close to water bodies.

The function will then try to find the difference between the different networks, and estimate the individual bias for each network, relative to a reference value, usually the average of all networks. The method is not recommended if there can be assumed to be a dependency beteween the process and the networks.

removeLocalBias removes the bias estimated in findLocalBias.

Value

From findLocalBias: A list consisting of one element for each regional network, or an element single if only one regional network is apparent. Each of these elements is again a list consisting of several other elements, where bias is the interesting one. The remaining elements are only necessary for debugging purposes. The elements D, V and Q refers to the matrices with same names in Skoien et al. (2009), i.e. the relationship matrix, the variance matrix and the difference matrix.

From removeLocalBias: A SpatialPointsDataFrame with the biases subtracted.

Author(s)

Jon Olav Skoien

References

Skoien, J. O., O. P. Baume, E. J. Pebesma, and G. B. M. Heuvelink. 2010. Identifying and removing heterogeneities between monitoring networks. Environmetrics 21(1), 66-84.

Examples

# Assuming that the soil type is the source of biases
data(meuse)
coordinates(meuse) = ~x+y


lb = findLocalBias(meuse,gid = "soil",formulaString=as.formula(zinc~1))
lb$single$bias

meuseUnbias = removeLocalBias(meuse,localBias = lb, gid = "soil",
    formulaString = zinc~1)