README.md
In ammaciel/stilf: Land Use Change Calculus

This package implements the LUC Calculus for reasoning about land use change events. Based on a set of classified time series, we build expressions to answer specific questions, such as Which events of "Forest" areas were replaced by "Pasture"?

With package "lucC" is possible to build questions using Allen's interval temporal logic relationships and also others extended from their study. I suggest the reader read (Allen 1983) and (Allen 1984) for more details. Besides, is possible to generate graphics with event information and plot maps with results. Using these events the user can to perform analysis on time series data to discover important land use changes.

wtss is an R client package for handling Web Time-Series Service (WTSS) in the client side at https://github.com/e-sensing/wtss.R

Classification method can be found using package dtwSat at https://github.com/vwmaus/dtwSat

Tools to Satellite Image Time Series analysis can be found using package sits at https://github.com/gilbertocamara/sits

Git
R
Rstudio
A set of time series classified data
The lucC requires "devtools" package is available.

Open RStudio
Install devtools install.packages("devtools")
Load devtools library(devtools)
Install the lucC package install_github("ammaciel/lucC")

Load the lucC package library(lucC)
Load a example data data("example_TWDTW")
Create new variable tibble format and apply lucC_standard_date_events to standardize start_date and end_date columns

alter start_date and end_date to a especific range in order to extract events

lucC_starting_point() data_tb <- example_TWDTW %>% lucC_standard_date_events(data_tb = ., month_year = "09", day_month = "01") %>% dplyr::select(longitude, latitude, start_date, end_date, label, id, index) data_tb

Plot example_1.tb lucC_plot_maps_input(data_tb, EPSG_WGS84 = TRUE, custom_palette = TRUE, RGB_color = c( "#FFB266", "#1b791f", "#929e6e", "#f5e7a1"))

Fig. 1. Plot time series classified data

Apply lucC_predicate_holds function to discover events of Pasture. For this is necessary create a for loop to read all data and not only one pixel over time

# p = properties of locations :
p1 <- "Pasture"

# t = interval:
t1 <- lucC_interval("2000-09-01","2017-03-01")

# Test holds for many time series 
question_holds <- function(data.tb){

  aux.df = NULL
  if (nrow(event2 <- lucC_predicate_holds(data.tb, p1, t1)) >= 1)
    aux.df <- event2
  else
    aux.df <- NULL

  data.frame(aux.df)

}

temp.tb <- data_tb

output.tb = data.frame(do.call("rbind", parallel::mclapply( X = split(temp.tb, temp.tb$index), 
                                                            mc.cores=1, #parallel::detectCores(),
                                                            FUN = question_holds)))
output.tb

See example_1.tb plot with lucC_plot_maps_input function lucC_plot_maps_input(data_tb, EPSG_WGS84 = TRUE, custom_palette = TRUE, RGB_color = c("#FFB266", "#1b791f", "#929e6e","#f5e7a1"))
See all events with lucC_plot_maps_events function lucC_plot_maps_events(output.tb, EPSG_WGS84 = TRUE, custom_palette = TRUE, RGB_color = c("#FFB266", "#1b791f", "#929e6e","#f5e7a1"), shape_point = 0, colour_point = "black", size_point = 2.3)

Fig. 2. Plot events discovered from time series classified data

See barplot total area in square kilometers lucC_plot_bar_events(output.tb, custom_palette = TRUE, RGB_color = "#929e6e", pixel_resolution = 231.6564, side_by_side = FALSE) and sequence plot lucC_plot_sequence_events(output.tb, end_date = "2017-03-01", custom_palette = TRUE, RGB_color = "#929e6e", show_y_index = FALSE)

Fig. 3.(a) Barplot with total area

Fig. 3.(b) Sequence plot

Apply lucC_predicate_holds function to discover events for only one pixel with events of Forest and Pasture.

# load library
library(lucC)
lucC_starting_point()

# load data example
data("example_TWDTW")
example_TWDTW

# alter start_date and end_date to a especific range in order to extract events
# because we work with annual intervals
# select only one time serie with index equals 13
data_tb <- example_TWDTW %>% 
  lucC_standard_date_events(data_tb = ., month_year = "09", day_month = "01") %>% 
  dplyr::select(longitude, latitude, start_date, end_date, label, id, index) %>% 
  dplyr::filter(., .$index == 13)
data_tb

# p = properties of locations :
p1 <- "Forest"
p2 <- "Pasture"

# t = interval:
t1 <- lucC_interval("2000-09-01","2004-09-01")
t2 <- lucC_interval("2004-09-01","2017-09-01")

# Test holds for one time serie
question_holds <- function(data.tb){

  aux.df = NULL

  if (nrow(ev1 <- lucC_predicate_holds(data.tb, p1, t1)) >= 1 &
      nrow(ev2 <- lucC_predicate_holds(data.tb, p2, t2)) >= 1 ){

    if(nrow(ev1) != 0 & nrow(ev2) != 0){

      if(isTRUE(lucC_relation_meets(tail(lucC_interval(ev1$start_date, ev1$end_date), 1),
                                    head(lucC_interval(ev2$start_date, ev2$end_date), 1))))
        aux.df <- rbind(ev1,ev2)
    } else 
      aux.df <- NULL

  } else {
    aux.df <- NULL
  }

  data.frame(aux.df)
}

temp.tb <- data_tb
output.tb2 = data.frame(do.call("rbind", parallel::mclapply( X = split(temp.tb, temp.tb$index), 
                                                             mc.cores=1, #parallel::detectCores(),
                                                             FUN = question_holds)))
output.tb2

View example_2.tb plot with lucC_plot_maps_input function lucC_plot_maps_input(data_tb, EPSG_WGS84 = TRUE, custom_palette = TRUE, RGB_color = c("#FFB266", "#1b791f", "#929e6e"))
View all events with lucC_plot_maps_events function lucC_plot_maps_events(output.tb2, EPSG_WGS84 = TRUE, custom_palette = TRUE, RGB_color = c("#FFB266", "#1b791f", "#929e6e"), shape_point = 4, colour_point = "blue", size_point = 8)

Fig. 4.(a) Pixel over time

Fig. 4.(b) Pixel with events

Apply lucC_predicate_holds function to discover for a sequence of events of Forest, Pasture, Single cropping and Double cropping in this order.

# load library
library(lucC)
lucC_starting_point()

# load data example
data("example_TWDTW")
example_TWDTW

# alter start_date and end_date to a especific range in order to extract events
# because we work with annual intervals
data_tb <- example_TWDTW %>% 
  lucC_standard_date_events(data_tb = ., month_year = "09", day_month = "01") %>% 
  dplyr::select(longitude, latitude, start_date, end_date, label, id, index) 
data_tb

# p = properties of locations :
p1 <- c("Forest", "Pasture", "Single_cropping", "Double_cropping")

# t = interval:
t1 <- lucC_interval("2000-09-01","2017-09-01")

# Test holds for one time serie
question_holds <- function(data.tb){

  aux.df = NULL

  aux.df <- lucC_event_transitions(data.tb, properties = p1, time_intervals = t1)

  data.frame(aux.df)
}

temp.tb <- data_tb
output.tb3 = data.frame(do.call("rbind", parallel::mclapply( X = split(temp.tb, temp.tb$index), 
                                                             mc.cores=1, #parallel::detectCores(),
                                                             FUN = question_holds)))
output.tb3

See plot with lucC_plot_maps_input function lucC_plot_maps_input(data_tb, EPSG_WGS84 = TRUE, custom_palette = TRUE, RGB_color = c("#FFB266", "#1b791f", "#929e6e","#f5e7a1"))
View all events with lucC_plot_maps_events function lucC_plot_maps_events(output.tb3, EPSG_WGS84 = TRUE, custom_palette = TRUE, RGB_color = c("#FFB266", "#1b791f", "#929e6e","#f5e7a1"), shape_point = 0, colour_point = "blue", size_point = 2.3)

Fig. 5. Plot events discovered from time series classified data

See barplot total area in square kilometers lucC_plot_bar_events(output.tb3, custom_palette = TRUE, RGB_color = c("#FFB266", "#1b791f", "#929e6e","#f5e7a1"), pixel_resolution = 250, side_by_side = FALSE) and sequence plot lucC_plot_sequence_events(output.tb3, show_y_index = FALSE, end_date = "2017-03-01", custom_palette = TRUE, RGB_color = c("#FFB266", "#1b791f", "#929e6e","#f5e7a1"), relabel = FALSE, original_labels = c("Double_cropping", "Forest", "Pasture", "Single_cropping"), new_labels = c("DC","F","P","SC"))