Satellite Geolocation for Animal Tracking

Description Usage Arguments Details Value

Fit an Essie model

1 2	essie(model, grid, epsilon1 = 0.001, epsilon2 = 1e-16, verbose = interactive())

`model`	a model structure generated by `essieThresholdModel`.
`grid`	a raster object defining the grid
`epsilon1`	likelihood tolerance
`epsilon2`	transition probability tolerance.
`verbose`	report progress at prompt?

Essie is a discrete analog of Stella that only considers locations on a lattice of grid points.

The posterior probability that the tag is at a particular location is determined by a two pass recursive algorithm. The forward sweep propogates location information forward in time, the backward sweep propogates location information backward in time, and the full posterior is a compromise of these two.

The method is only approximate and is controlled by the two tolerances epsilon1 and epsilon2. The first limits the precision of the likelihood - locations for which the likelihood fall below epsilon1 of its maximum are ignored. The second limits the precision of the transition probabilities - locations for which the probability below epsilon2 of the maximum are ignored in the movement calculation. The smaller these parameters are set, the slower but more accurate the fit.

a list with elements

`grid`	a raster object that defines the grid.
`times`	the times corresponding to the location estimates.
`lattice`	a sparse grid representation of the posterior location probabilities.