Hi Kevin- You have numerous options. I'll list a few as function(package) format for R packages, and there are more options for which a search through 'Environmetrics' Taskview in R would probably reveal.
approx(stats) # linear interpolation spline(stats) # spline interpolation crwPredict(crawl) # state-space interpolation parameterized by measurement errors (GPS errors) Depending on how in depth you want to evaluate the best methods, you could approach this by simulating some CRW pathways based on parameters from your real data (turning and velocity distributions). From those simulations you could randomly remove some points to reflect the frequency of your missing data, and then test which interpolation method gives the most desirable predictions of observations you've removed. Hope this helps, Tim Sippel - Postdoctoral researcher University of Hawaii On Wed, Feb 2, 2011 at 10:31 AM, Middel, Kevin (MNR) < kevin.mid...@ontario.ca> wrote: > Hi List, > > > > I’ve been looking for reasonable methods of estimating missing GPS tracking > data to fill in gaps at a regular time interval. For instance if I have > data being collected every 4 hours, but have a 12 hour gap between a set of > locations, is it possible to interpolate the missing 3 locations between > point A and B? I have been experimenting with some of the simm.* functions > in adehabitat, but am wondering if anyone has any experience with methods > that might use the fractal dimension of the path as an estimation parameter. > I’m still trying to get my head around this, but am thinking that if I know > the Fd of the path, then there may be a way to use that as a parameter to > estimate the missing points. I’m hoping that using the Fd of the path would > presumably provide a more ‘realistic’ estimation of the path than a straight > CRW or Brownian estimate (but I may be way off with this). > > > > I’ve been looking online and through the literature for any insight, but > haven’t found much so far. If anyone has some experience with this or some > other ideas I’d really appreciate hearing from you. > > > > Thanks very much, > > > > Kevin Middel > > M.Sc. Candidate, > > Trent University > > > > _______________________________________________ > AniMov mailing list > AniMov@faunalia.it > http://lists.faunalia.it/cgi-bin/mailman/listinfo/animov > >
_______________________________________________ AniMov mailing list AniMov@faunalia.it http://lists.faunalia.it/cgi-bin/mailman/listinfo/animov
