random_steps producing steps far outside data extent

[camkie27]

Hi there,

I am attempting to conduct iSSF on data and am hitting a roadblock when trying to produce random steps. My data are in EPSG 4326, with long/lat coordinates usually beginning with -97.xxx and 26.xxx, respectively. I've checked that my crs of all of my objects remains consistent up to the point of using random_steps, and then the random step coordinates begin with numbers nowhere close to my data - some beginning with -108 or 15, for example. I've included sample code below that reproduces my problem. I have amt version ‘0.3.0.0’

#example data:
location.long=c(-97.47371, -97.47388, -97.47392, -97.47395, -97.47394, -97.47413, -97.47411, -97.47414)
location.lat=c(26.52136, 26.52141, 26.52137, 26.52169, 26.52171, 26.52156, 26.52159, 26.52156)
timestamp=c("2022-06-25 00:00:00", "2022-06-25 05:00:00", "2022-06-25 10:00:00", "2022-06-25 15:00:00", "2022-06-30 20:00:00", "2022-07-01 01:00:00", "2022-07-01 06:00:00", "2022-07-01 11:00:00")
timestamp<-as.POSIXct(timestamp, format = "%Y-%m-%d %H:%M:%S")

samp<-data.frame(location.long, location.lat, timestamp)

#make track and steps and random steps
samp.track<-samp %>% 
  dplyr::select(x = 'location.long', y = 'location.lat', t = 'timestamp') %>%
  make_track(x , y , t, crs = 4326) %>% track_resample(rate = hours(5), tolerance = minutes(0)) 

samp.steps<- samp.track %>%
  steps_by_burst(lonlat = TRUE) 

get_crs(samp.steps) #still in the proper crs

rand.steps<-samp.steps %>%
  mutate(sl_ = ifelse(sl_ == 0, 0.01, sl_)) %>% 
  random_steps(n_control=2)

rand.steps

The output of the above code with random steps:

I've tried researching the issue and haven't found anything. Am I missing something?

Thanks in advance,
c

[jmsigner]

I am not sure, that I fully understand the problem.

library(amt)
#> 
#> Attaching package: 'amt'
#> The following object is masked from 'package:stats':
#> 
#>     filter
  library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
set.seed(3)

#example data:
location.long=c(-97.47371, -97.47388, -97.47392, -97.47395, -97.47394, -97.47413, -97.47411, -97.47414)
location.lat=c(26.52136, 26.52141, 26.52137, 26.52169, 26.52171, 26.52156, 26.52159, 26.52156)
timestamp=c("2022-06-25 00:00:00", "2022-06-25 05:00:00", "2022-06-25 10:00:00", "2022-06-25 15:00:00", "2022-06-30 20:00:00", "2022-07-01 01:00:00", "2022-07-01 06:00:00", "2022-07-01 11:00:00")
timestamp<-as.POSIXct(timestamp, format = "%Y-%m-%d %H:%M:%S")

samp<-data.frame(location.long, location.lat, timestamp)

#make track and steps and random steps
samp.track<-samp %>% 
  dplyr::select(x = 'location.long', y = 'location.lat', t = 'timestamp') %>%
  make_track(x , y , t, crs = 4326) %>% track_resample(rate = hours(5), tolerance = minutes(0)) 

samp.steps<- samp.track %>%
  steps_by_burst(lonlat = TRUE) 

get_crs(samp.steps) #still in the proper crs
#> Coordinate Reference System:
#>   User input: EPSG:4326 
#>   wkt:
#> GEOGCRS["WGS 84",
#>     ENSEMBLE["World Geodetic System 1984 ensemble",
#>         MEMBER["World Geodetic System 1984 (Transit)"],
#>         MEMBER["World Geodetic System 1984 (G730)"],
#>         MEMBER["World Geodetic System 1984 (G873)"],
#>         MEMBER["World Geodetic System 1984 (G1150)"],
#>         MEMBER["World Geodetic System 1984 (G1674)"],
#>         MEMBER["World Geodetic System 1984 (G1762)"],
#>         MEMBER["World Geodetic System 1984 (G2139)"],
#>         ELLIPSOID["WGS 84",6378137,298.257223563,
#>             LENGTHUNIT["metre",1]],
#>         ENSEMBLEACCURACY[2.0]],
#>     PRIMEM["Greenwich",0,
#>         ANGLEUNIT["degree",0.0174532925199433]],
#>     CS[ellipsoidal,2],
#>         AXIS["geodetic latitude (Lat)",north,
#>             ORDER[1],
#>             ANGLEUNIT["degree",0.0174532925199433]],
#>         AXIS["geodetic longitude (Lon)",east,
#>             ORDER[2],
#>             ANGLEUNIT["degree",0.0174532925199433]],
#>     USAGE[
#>         SCOPE["Horizontal component of 3D system."],
#>         AREA["World."],
#>         BBOX[-90,-180,90,180]],
#>     ID["EPSG",4326]]

rand.steps<-samp.steps %>%
  mutate(sl_ = ifelse(sl_ == 0, 0.0001, sl_)) %>% 
  random_steps(n_control=2)

rand.steps
#> # A tibble: 12 × 12
#>    burst_   x1_    x2_   y1_   y2_   sl_      ta_ t1_                
#>  *  <dbl> <dbl>  <dbl> <dbl> <dbl> <dbl>    <dbl> <dttm>             
#>  1      1 -97.5  -97.5  26.5 26.5   5.97 -1.15    2022-06-25 05:00:00
#>  2      1 -97.5  -89.4  26.5 24.8   8.26 -3.06    2022-06-25 05:00:00
#>  3      1 -97.5  -92.7  26.5 20.2   7.89  2.51    2022-06-25 05:00:00
#>  4      1 -97.5  -97.5  26.5 26.5  35.7   2.32    2022-06-25 10:00:00
#>  5      1 -97.5 -111.   26.5 13.5  18.5  -0.00730 2022-06-25 10:00:00
#>  6      1 -97.5  -93.2  26.5 17.5   9.95  1.23    2022-06-25 10:00:00
#>  7      2 -97.5  -97.5  26.5 26.5   3.88  2.83    2022-07-01 01:00:00
#>  8      2 -97.5 -114.   26.5  3.15 28.6   0.291   2022-07-01 01:00:00
#>  9      2 -97.5  -95.8  26.5 16.2  10.4   1.06    2022-07-01 01:00:00
#> 10      2 -97.5  -97.5  26.5 26.5   4.48 -2.95    2022-07-01 06:00:00
#> 11      2 -97.5  -75.6  26.5 24.4  22.0  -1.08    2022-07-01 06:00:00
#> 12      2 -97.5  -97.5  26.5 30.3   3.77  0.602   2022-07-01 06:00:00
#> # ℹ 4 more variables: t2_ <dttm>, dt_ <drtn>, case_ <lgl>, step_id_ <dbl>

table(rand.steps$x1_) 
#> 
#> -97.47413 -97.47411 -97.47392 -97.47388 
#>         3         3         3         3
table(rand.steps$y1_) 
#> 
#> 26.52137 26.52141 26.52156 26.52159 
#>        3        3        3        3

table(rand.steps$x2_) 
#> 
#> -113.879248634899 -110.669223439406 -97.5283911944845         -97.47414 
#>                 1                 1                 1                 1 
#>         -97.47411         -97.47395         -97.47392 -95.8267003725921 
#>                 1                 1                 1                 1 
#> -93.2095752976797  -92.689222693138 -89.3983992952093 -75.5721597557808 
#>                 1                 1                 1                 1
table(rand.steps$y2_)
#> 
#> 3.14614656780088 13.5174193850682 16.2348580171546 17.5306353076121 
#>                1                1                1                1 
#> 20.2474726434914 24.4109954510294 24.8055718625155         26.52137 
#>                1                1                1                1 
#>         26.52156         26.52159         26.52169 30.2881405405134 
#>                1                1                1                1

For each starting point (x1_ and y1_) you have three times the same coordinate (observed + random steps) and all the end points are different. This is what you would expect? Do I miss something?

^{Created on 2024-11-26 with reprex v2.1.0}

[camkie27]

That part seems fine, yes.

My main question is why are the coordinates of some of the random steps (x2, y2) in drastically different places? They should all start with -97 and 26.

[jmsigner]

The starting points of the random steps should be the same as for the observed steps (column x1_ and y1_). The end points should be different. Step lengths (sl_) seem to be way too large, I will look into this.

[camkie27]

Thank you so much!

[camkie27]

Just as another note - I expect the units for sl_ to be meters, and the values themselves in the table above seem reasonable for my study species (turtles). I've mapped used steps in ArcGIS and confirmed that the sl_ distance between them is in fact in meters and is accurately measured.

Thank you again for your time and efforts looking into this - I really appreciate it.

[RonEfrat]

Hi, jumping in here because I have similar problems. Is seems that the sl_ is OK (the values seem to be taken from the right distribution), but there's no turn angle that will reach x2_ y2_ given the simulated step lengths.
For example, if you take the fifth row in the above example, the latitudes you can get with a step length of 42 km are between ca. 26.1 and 26.9, and the longitudes between ca. -97.9 and -97.
My guess is that the calculation of destination done to get x2 and y2 has some issues.

[camkie27]

Hi again,

As a last resort, I tried adding the following line of code after the make_track function to project my data to my UTM zone rather than using EPSG 4326, and it produced viable steps!

transform_coords(crs_to = 32614, crs_from = 4326)

Seems as though the issue lies within using crs EPSG 4326...

Thanks @jmsigner for all your hard work and time! I am eternally grateful for this package and couldn't do my dissertation without it.

Best,
c

[jmsigner]

Sorry for the delay. I think I found the bug. Could you try again and replace

samp.steps<- samp.track %>%
  steps_by_burst(lonlat = TRUE) 

with

samp.steps<- samp.track %>%
  steps_by_burst()