Camera data wrangling
camera-data-wrangling.RmdThe use of camera traps in ecological studies has become increasingly popular for monitoring wildlife. Managing and analyzing camera trap data efficiently is important for extracting meaningful and accurate insights. In this vignette, we will explore how to perform camera data wrangling, specifically focusing on the ABMI’s Ecosystem Health 2014 data set.
Use the wt_download_report() function to retrieve the
main report for the CAM sensor in the Ecosystem Health 2014 project:
eh14_raw <- wt_download_report(
project_id = 205,
sensor_id = "CAM",
report = "main"
)
head(eh14_raw)## project_id location location_id latitude longitude location_buffer_m
## 1 205 509-NW 47752 56.6338 -111.7007 5500
## 2 205 509-NW 47752 56.6338 -111.7007 5500
## 3 205 509-NW 47752 56.6338 -111.7007 5500
## 4 205 509-NW 47752 56.6338 -111.7007 5500
## 5 205 509-NW 47752 56.6338 -111.7007 5500
## 6 205 509-NW 47752 56.6338 -111.7007 5500
## equipment_serial image_id image_date_time image_set_id image_fov
## 1 P900HF12171487 5584838 2014-03-24 14:05:41 1344 <NA>
## 2 P900HF12171487 5584839 2014-03-24 14:05:43 1344 <NA>
## 3 P900HF12171487 15413607 2014-03-24 14:05:45 1344 <NA>
## 4 P900HF12171487 15413608 2014-03-24 14:05:46 1344 <NA>
## 5 P900HF12171487 15413500 2014-03-24 14:05:47 1344 <NA>
## 6 P900HF12171487 15413561 2014-03-24 14:05:50 1344 <NA>
## image_snow image_snow_depth_m image_water_depth_m species_scientific_name
## 1 NA NA NA <NA>
## 2 NA NA NA <NA>
## 3 NA NA NA <NA>
## 4 NA NA NA <NA>
## 5 NA NA NA <NA>
## 6 NA NA NA <NA>
## species_common_name individual_count age_class sex_class behaviours
## 1 STAFF/SETUP VNA VNA VNA <NA>
## 2 STAFF/SETUP VNA VNA VNA <NA>
## 3 STAFF/SETUP VNA VNA VNA <NA>
## 4 STAFF/SETUP VNA VNA VNA <NA>
## 5 STAFF/SETUP VNA VNA VNA <NA>
## 6 STAFF/SETUP VNA VNA VNA <NA>
## health_diseases coat_colours coat_attributes tine_attributes direction_travel
## 1 <NA> <NA> <NA> <NA> <NA>
## 2 <NA> <NA> <NA> <NA> <NA>
## 3 <NA> <NA> <NA> <NA> <NA>
## 4 <NA> <NA> <NA> <NA> <NA>
## 5 <NA> <NA> <NA> <NA> <NA>
## 6 <NA> <NA> <NA> <NA> <NA>
## has_collar has_eartag ihf observer observer_id tag_comments
## 1 NA NA <NA> Kate Broadley 245 <NA>
## 2 NA NA <NA> Kate Broadley 245 <NA>
## 3 NA NA <NA> Kate Broadley 245 <NA>
## 4 NA NA <NA> Kate Broadley 245 <NA>
## 5 NA NA <NA> Kate Broadley 245 <NA>
## 6 NA NA <NA> Kate Broadley 245 <NA>
## tag_needs_review tag_is_verified image_in_wildtrax tag_id
## 1 FALSE FALSE TRUE 3368979
## 2 FALSE FALSE TRUE 3368980
## 3 FALSE FALSE TRUE 9246962
## 4 FALSE FALSE TRUE 9246963
## 5 FALSE FALSE TRUE 9246918
## 6 FALSE FALSE TRUE 9246946Evaluate independent detections with
wt_ind_detect():
# Back to the Ecosystem Health 2014 data.
eh14_detections <- wt_ind_detect(
x = eh14_raw,
threshold = 30,
units = "minutes",
remove_human = TRUE,
remove_domestic = TRUE
)
head(eh14_detections, width = 75) ## detection project_id location
## 1 Snowshoe Hare 1 205 509-NW
## 2 White-tailed Deer 1 205 509-NW
## 3 Black Bear 1 205 509-SW
## 4 Canada Lynx 1 205 509-SW
## 5 Gray Wolf 1 205 509-SW
## 6 Grouse, Ptarmigan and allies 1 205 509-SW
## species_common_name start_time end_time
## 1 Snowshoe Hare 2014-03-27 02:10:28 2014-03-27 02:10:28
## 2 White-tailed Deer 2014-06-20 11:35:08 2014-06-20 11:35:08
## 3 Black Bear 2014-06-05 07:30:17 2014-06-05 07:33:38
## 4 Canada Lynx 2014-04-04 22:19:04 2014-04-04 22:20:04
## 5 Gray Wolf 2014-04-28 19:04:00 2014-04-28 19:06:57
## 6 Grouse, Ptarmigan and allies 2014-03-25 13:39:12 2014-03-25 13:39:16
## total_duration_seconds n_images avg_animals_per_image max_animals
## 1 0 1 1 1
## 2 0 1 1 1
## 3 201 9 1 1
## 4 60 9 1 1
## 5 177 23 1 1
## 6 4 3 1 1So there are 313 independent detections in this data set, when using a threshold of 30 minutes.
The output from wt_ind_detect() gave us some useful
information. But we probably need to do additional wrangling for our
data to be in the proper format for certain modeling techniques
(e.g. habitat modeling, occupancy). For example, we want to evaluate the
number of detections in a specified time interval (e.g. daily, weekly,
or monthly), including zeroes.
Summarise your camera data
With wt_summarise_cam() you can get:
- The output from
wt_ind_detect()(e.g. the objecteh14_detections) - Your raw data (e.g. the object
eh14_raw) - The time interval you’re interested in (e.g. weekly)
- The variable you’re interested in (e.g. detections, presence/absence)
- The desired output format (‘wide’ or ‘long’)
# A call to `wt_summarise_cam()`:
eh14_summarised <- wt_summarise_cam(
# Supply your detection data
detect_data = eh14_detections,
# Supply your raw image data
raw_data = eh14_raw,
# Now specify the time interval you're interested in
time_interval = "week",
# What variable are you interested in?
variable = "detections",
# Your desired output format (wide or long)
output_format = "wide"
)
head(eh14_summarised)## project_id location year week n_days_effort Black Bear Canada Lynx Coyote
## 1 205 509-NW 2014 13 7 0 0 0
## 2 205 509-NW 2014 14 7 0 0 0
## 3 205 509-NW 2014 15 7 0 0 0
## 4 205 509-NW 2014 16 7 0 0 0
## 5 205 509-NW 2014 17 7 0 0 0
## 6 205 509-NW 2014 18 7 0 0 0
## Cranes, Rails, Coots Crow Deer Ducks, Swans, Geese Fisher Gray Jay Gray Wolf
## 1 0 0 0 0 0 0 0
## 2 0 0 0 0 0 0 0
## 3 0 0 0 0 0 0 0
## 4 0 0 0 0 0 0 0
## 5 0 0 0 0 0 0 0
## 6 0 0 0 0 0 0 0
## Grouse, Ptarmigan and allies Marten Moose Red Fox Red Squirrel Snowshoe Hare
## 1 0 0 0 0 0 1
## 2 0 0 0 0 0 1
## 3 0 0 0 0 0 0
## 4 0 0 0 0 0 0
## 5 0 0 0 0 0 0
## 6 0 0 0 0 0 0
## Songbird Unidentified White-tailed Deer
## 1 0 0 0
## 2 0 0 0
## 3 0 0 0
## 4 0 0 0
## 5 0 0 0
## 6 0 0 0And now you can get straight into the science!