In sustainablefarms/sflddata: Data Extraction Methods for the Linking Data Project

knitr::opts_chunk$set(echo = TRUE)
out <- lapply(c("sf", "tsibble", 'lubridate', "viridis",
                'ggplot2', 'tidyr', 'grid', 'gridExtra', 
                'feasts', 'dplyr', 'gtable', 'fable',
                'mgcv', 'readxl'),
       library, character.only = TRUE)

Import Data

graz_farms <- read_excel("./private/data/raw/GrazingStudy_GrazingRecordData.xlsx", sheet = "Farm")
graz_siteinfo <- read_excel("../private/data/raw/GrazingStudy_GrazingRecordData.xlsx", sheet = "Site") %>%
  mutate_at(vars(`Govt Stewardship zone`, Treatment, `Bark Study Site`,
                 Catchment, `LLS Boundary`, `Original Stewardship Study`,
                 `Steward/Control Site`, `Background`, `Treatment Code`,
                 Aspect, `Rock Type`, `Vegetation Formation`, `Vegetation Type`
                 ),
            as.factor) %>%
  mutate_at(vars(Asp_N, Asp_E),
            as.numeric) 
graz_grazinginfo <- read_excel("../private/data/raw/GrazingStudy_GrazingRecordData.xlsx", sheet = "Grazing") %>%
  mutate_at(vars(Event, `Spring Grazed`, `Summer Grazed`,
                 `Autumn Grazed`, `Winter Grazed`
                 ),
            as.factor) %>%
  mutate_at(vars( Year, `Farmer DSE`),
            as.numeric) %>%
  select_if(.predicate = function(x) sum(is.na(x)) != nrow(grazing)) #delete empty columns

Summary by Column

dim(graz_grazinginfo)
colnames(graz_grazinginfo)
summary(graz_grazinginfo)

dim(graz_siteinfo)
colnames(graz_siteinfo)
summary(graz_siteinfo)

Derived Columns

graz_grazinginfo <- grazing %>%
  mutate(`Event Duration` = as.duration(difftime(`Event Finish`, `Event Start`, units = "days")))

Spatial Figures

graz_grazinginfo %>%
  filter(!is.na(Event)) %>%
  filter(Event %in% c("Destocked", "Grazed")) %>% #basically all the events
  group_by(`Site Code`, Event) %>%
  # select(Size_ha, `Event Duration`) %>%
  summarize(tot.time = sum(`Event Duration`)) %>%
  pivot_wider(id_cols = `Site Code`, names_from = Event, values_from = tot.time) %>%
  mutate(Destocked.prop = Destocked / (Grazed + Destocked)) %>%
  inner_join(graz_siteinfo, by = "Site Code") %>%
  ggplot() +
  geom_point(aes(x = `LONG 0m`, y = `LAT 0m`, col = Destocked.prop,
                 shape = is.na(Destocked.prop))) +
  scale_color_viridis() +
  # scale_size_area(max_size = 2) +
  coord_fixed() +
  xlab("Longitude") + ylab("Latitude") +
  ggtitle("Proportion of Time Spent Destocked")