calc_spatial_stats: Calculate spatial statistics
In cbuelo/tvsews: Time vs. Space Early Warning Statistics

Description Usage Arguments Details Value Examples

Calculates spatial early warning statistics (spatial standard deviation and spatial autocorrelation = Moran's I) from a data frame of (optionally multiple) sampling events and variables. By default uses multiple cores to speed up spatial autocorrelation calculation, which can take quite a bit of time depending on how many data points you have.

calc_spatial_stats(
  spatial_data = flame_data,
  lat_col = "latitude",
  lon_col = "longitude",
  var_cols = c("BGApc_ugL_tau", "ODO_percent_tau", "pH_tau"),
  id_cols = c("Lake", "Year", "DOY"),
  statistics = c("SD", "moransI"),
  multiple_cores = TRUE
)

`spatial_data`	data frame containing spatial locations and measurements of one or more variables; see `flame_data` for default and formatting
`lat_col`	character string, column name for latitude column
`lon_col`	character string, column name for longitude column
`var_cols`	character vector, column names that hold measurements of different variables
`id_cols`	character vector, columns that identify unique sampling events to separate data into before calculating stats independently on, defaults are "Lake", "Year", and "DOY"
`statistics`	character vector, statistics to run; options are "SD", "moransI", "skew", and "kurt"
`multiple_cores`	TRUE (default) or FALSE, should spatial autocorrelation calculations be run on multiple cores to speed up calculations?

Note that time estimates printed out are very rough estimates and are based on using the default data and variables, and run on a AMD Ryzen 1800X with 8 cores/16 threads.

data frame with calculated spatial stats (SD and Moran's I) for each sample and variable

spat_stats <- calc_spatial_stats(flame_data)
library(ggplot2)
ggplot(spat_stats %>% dplyr::filter(Stat == "SD"), aes(x = DOY, y = Value, color = Lake)) +
  geom_line() +
  facet_grid(rows = vars(Variable), cols = vars(Year), scales = "free_y") +
  theme_bw()