RooksNeighCt: Define a subset of grid locations with non-overlapping rook...

In ICvectorfields: Vector Fields from Spatial Time Series of Population Abundance

Define a subset of grid locations with non-overlapping rook neighborhoods

Description

This function prunes the data frame returned by the PatternDetect function such that it includes only rook's neighborhoods that do not overlap. Pruning is done by sequential removal of observations that are too near one another and as a result are overlapping. Locations that are the most highly connected are removed first.

Usage

RooksNeighCt(vfdf)

Arguments

vfdf	A data frame as returned by PatternDetect

Details

The reason for this function's existence is to facilitate probabilistic calculations regarding whether certain patterns are occurring more or less often that would be expected by chance. If rook's neighborhoods in which patterns are observed overlap, then the assumption of probabilistic independence is necessarily incorrect. Thus, overlap invalidates any calculation of the probability of occurrence of a particular pattern if that calculation assumes independence. The pruning actions of this function enable the user to more safely assume probabilistic independence.

Value

A data frame similar to vfdf except that it includes only grid locations with speed estimates in all four adjacent grid locations in their rook's neighborhood. An additional column called IndPatternCt is appended which contains NA values for locations that are overlapping other locations and ones for all non-overlapping locations that have speed estimates in all four adjacent cells.

Examples

# creating convergence/divergence patterns
Mat1 <- matrix(rep(0,9*9), nrow = 9)
Mat1[3, c(4, 6)] <- 1
Mat1[7, c(4, 6)] <- 1
Mat1[c(4, 6), 3] <- 1
Mat1[c(4, 6), 7] <- 1
Mat1

Mat2 <- matrix(rep(0,9*9), nrow = 9)
Mat2[2, c(4, 6)] <- 1
Mat2[8, c(4, 6)] <- 1
Mat2[c(4, 6), 2] <- 1
Mat2[c(4, 6), 8] <- 1
Mat2

Mat1 <- cbind(Mat1, Mat1)
Mat1 <- rbind(Mat1, Mat1)

Mat2 <- cbind(Mat2, Mat2)
Mat2 <- rbind(Mat2, Mat2)

# rasterizing
rast1 <- terra::rast(Mat1)
terra::plot(rast1)
rast2 <- terra::rast(Mat2)
terra::plot(rast2)

# Detecting a divergence
(VFdf1 <- DispField(rast1, rast2, factv1 = 3, facth1 = 3, restricted = TRUE))
(patdf1 <- PatternDetect(VFdf1))
(subdf1 <- RooksNeighCt(patdf1))
# The last call should print a data table with four rows, each with a one under
# the column header of IndPatternCt

ICvectorfields documentation built on March 18, 2022, 7:34 p.m.