ddfilter function

Filter locations using a data driven filter

Function to remove locations by a data driven filter as described in Shimada et al. (2012).

ddfilter(sdata, vmax = 8.9, vmaxlp = 1.8, qi = 4, ia = 90, method = 1)

Arguments

• sdata: A data frame containing columns with the following headers: "id", "DateTime", "lat", "lon", "qi". See the data turtle for an example. The function filters the input data by a unique "id" (e.g. transmitter number, identifier for each animal). "DateTime" is the GMT date & time of each location in class POSIXct or character with the following format "2012-06-03 01:33:46". "lat" and "lon" are the latitude and longitude of each location in decimal degrees. "qi" is the quality index associated with each location fix. The input values can be either the number of GPS satellites or Argos Location Classes. Argos Location Classes will be converted to numerical values, where "A", "B", "Z" will be replaced with "-1", "-2", "-3" respectively. The greater number indicates a higher accuracy.
• vmax: A numeric value specifying a threshold of speed from a previous and/or to a subsequent fix. Default is 8.9km/h. If this value is unknown, it can be estimated from sdata using the function vmax.
• vmaxlp: A numeric value specifying a threshold of speed, which is used to evaluate the locations of loop trips. Default is 1.8 km/h. If this value is unknown, it can be estimated from sdata using the function vmaxlp.
• qi: An integer specifying a threshold of quality index, which is used to evaluate the locations of loop trips. Default is 4.
• ia: An integer specifying a threshold of inner angle, which is used to evaluate the locations of loop trips. Default is 90 degrees.
• method: An integer (1 or 2) specifying how locations should be filtered with vmax. Default is 1 (both way) and removes a location if the speed from a previous AND to a subsequent location exceeds vmax. Select 2 (one way) to remove a location if the speed from a previous OR to a subsequent location exceeds vmax. For the latter, the filter examines successive suspect locations (i.e. the speed from a previous and/or to a subsequent location exceeds vmax) and retain one location that is associated with the minimum speed from a previous and/or to a subsequent location.

Returns

The input data is returned without locations identified by this filter. The following columns are added: "pTime", "sTime", "pDist", "sDist", "pSpeed", "sSpeed", "inAng". "pTime" and "sTime" are hours from a previous and to a subsequent fix respectively. "pDist" and "sDist" are straight distances in kilometres from a previous and to a subsequent fix respectively. "pSpeed" and "sSpeed" are linear speed from a previous and to a subsequent fix respectively. "inAng" is the degree between the bearings of lines joining successive location points.

Details

Locations are removed if the speed from a previous and/or to a subsequent location exceeds vmax, or if all of the following criteria apply: the associated quality index is less than or equal to qi, the inner angle is less than or equal to ia and the speed either from a previous or to a subsequent location exceeds vmaxlp. If vmax and vmaxlp are unknown, they can be estimated using the functions vmax and vmaxlp respectively.

Examples

#### Load data sets
## Fastloc GPS data obtained from a green turtle
data(turtle)

## A Map for the example site
data(Australia)
data(SandyStrait)

#### Filter temporal and/or spatial duplicates
turtle.dup <- dupfilter(turtle, step.time=5/60, step.dist=0.001)
 

#### ddfilter
## Using the built-in function to estimate the threshold speeds
V <- vmax(turtle.dup)
VLP <- vmaxlp(turtle.dup)
turtle.dd <- ddfilter(turtle.dup, vmax=V, vmaxlp=VLP)

## Or using user specified threshold speeds
turtle.dd <- ddfilter(turtle.dup, vmax=9.9, qi=4, ia=90, vmaxlp=2.0)

#### Plot data removed or retained by ddfilter
## Entire area
p1 <- to_map(turtle.dup, bgmap=Australia, point.size = 2, line.size = 0.5, axes.lab.size = 0, 
            multiplot = FALSE, point.bg = "red",
            title.size=15, title="Entire area")[[1]] + 
  geom_point(aes(x=lon, y=lat), data=turtle.dd, size=2, fill="yellow", shape=21)+
  geom_point(aes(x=x, y=y), data=data.frame(x=c(154, 154), y=c(-22, -22.5)), 
             size=3, fill=c("yellow", "red"), shape=21) + 
  annotate("text", x=c(154.3, 154.3), y=c(-22, -22.5), label=c("Retained", "Removed"), 
           colour="black", size=4, hjust = 0)

## Zoomed in
p2 <- to_map(turtle.dup, bgmap=SandyStrait, xlim=c(152.7, 153.2), ylim=(c(-25.75, -25.24)), 
            axes.lab.size = 0, point.size = 2, point.bg = "red", line.size = 0.5, 
            multiplot = FALSE, title.size=15, title="Zoomed in")[[1]] + 
geom_path(aes(x=lon, y=lat), data=turtle.dd, linewidth=0.5, colour="black", linetype=1) + 
geom_point(aes(x=lon, y=lat), data=turtle.dd, size=2, colour="black", shape=21, fill="yellow")

gridExtra::marrangeGrob(list(p1, p2), nrow=1, ncol=2)

References

Shimada T, Jones R, Limpus C, Hamann M (2012) Improving data retention and home range estimates by data-driven screening. Marine Ecology Progress Series 457:171-180 tools:::Rd_expr_doi("10.3354/meps09747")

See Also

ddfilter_speed, ddfilter_loop, vmax, vmaxlp

Author(s)

Takahiro Shimada