Irregular Sampling Schedules - CTMM variogram

[Guus]

Hi CTMM community,

I'm currently working with a GPS GSM dataset from Northern Lapwings (Vanellus vanellus), and I'm a bit struggling fitting the variogram. I found that my variogram has quite a spiky appearance, which probably is caused by the different frequency in sampling rate (in my dataset, during day time GPS locations were saved roughly every two minutes, while during nighttime only one point per hour was recorded. According to the variogram vignette (https://ctmm-initiative.github.io/ctmm/articles/variogram.html) I should use the dt argument within the variogram function, which I tried, but I still seem to have a large confidence interval. 

The R code I used for fitting the variogram:

# Set dt interval

 dt <- c(2, 60) %#% "minutes"
 
  # Calculate variogram
  vg <- variogram(c_data, dt = dt)
 
  # Fit variogram
  variogram.fit(vg) 

Am I missing a crucial step in the preprocessing? I would like to hear from you!! 

Cheers!

Guus

[Alex Brunswick]

Hi Guus,

You said that the relocations are saved roughly every 2 minutes, and hourly at night. Did you figure that out using the dt.plot function? There is info in the vignettes but it will let you visualise the time-lags between fixes. You can use this to more accurately decide on your dt argument in the variogram.

As far as my knowledge goes, an animal can still be a range resident if it’s variogram is ‘spiky’.

Hope this helps

Alex

--
You received this message because you are subscribed to the Google Groups "ctmm R user group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ctmm-user+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ctmm-user/ebf9523f-52ae-440b-957d-d3f219b5b03an%40googlegroups.com.

[Guus]

Hi Alex,

Thank you for your response! I figured the relocations interval by checking the timestamp column of my dataset. 

I will look into the vignette to visualise the time-lags between fixes!! 

Thanks!

Cheers,

Guus

Op vrijdag 30 januari 2026 om 01:04:22 UTC+1 schreef alexbrun...@gmail.com:

[Christen Fleming]

Hi Guus,

The default CIs aren't very accurate after the asymptote. CI="Gauss" is more accurate on data that are somewhat regularly sampled, but it is much slower to compute.

It looks like there is some mismatch at 60 minutes between the 2-minute data and the 1-hour data. This is likely because the 2-minute data considers only day movement and the 1-hour data considers both day and night movement. If the individuals don't move at night, you might just consider tossing out that data and doing something like dt=c(2%#%"min",1%#%"hr",12%#%"hr",24%#%"hr") .

Best,

Chris