search

positioning / kalmanfilter

PFRP and LPRC geolocation packages for archival tagged marine animals
http://positioning.github.io
12 stars 5 forks source link

longitude issue? #8

Open jflamarre opened 8 years ago

jflamarre commented 8 years ago

Hi,

I'm trying to use kftrack on a mac (El capitan) running Studio 0.99.903 and R 3.02.

All seems to run smoothly but longitudes look wrong and off the raw data, even int your example (big.241): kftrack_example big 241

See my results here with: fit<-kftrack(TEST, fix.first=FALSE, fix.last=FALSE, var.struct = "solstice")

kftrackamgp_k875

it seems that in my example, the longitude is off 30degree while latitude seems ok. Could you shed light on why this may be giving such results.

Thanks for such great and simple implementation of the Kalman filter. Cheers JF

tagtuna commented 8 years ago

Hi JF

The example, big241 is run correctly and the output looks good.

I don't have enough knowledge of what your own data was collected. But I would suggest, if you experiment with the following options (or combinations of such), and let's see what output the model would give. Make sure you turn on the CI=T to check too.

fit<-kftrack(TEST, fix.first=TRUE) # It's safe to assume you know where you tagged the animal. fit<-kftrack(TEST, fix.first=TRUE, fix.last=TRUE) # You likely have a good satellite fix without too much drift fit<-kftrack(TEST, bx.a=FALSE) # simplify the model, turn off longitude bias

Cheers, Tim