In RobinKohrs/rainfallR: Extract and manipulate gridded rainfall measurements

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(ggplot2)
library(purrr)
library(sf)
library(rainfallR)

Many times one wants to extract the rainfall for landslide locations (or non-landslide locations) that might not come from the IFFI database
The rainfallR pacakge was build only having the IFFI data in mind, so there are some quirks that I'll show here
First, lets create some data

Make the data

# get the shape of ST
st = iffitoR::get_shape_southtyrol() %>% st_transform(crs=32632)

# Get some random points
bb = st_bbox(st)
xs = sample(seq(bb[[1]], bb[[3]], length.out=100), 50)
ys = sample(seq(bb[[2]], bb[[4]], length.out=100), 50)

pts = map2_df(xs, ys, ~data.frame(x = .x, y=.y)) %>% 
  st_as_sf(coords = c("x", "y"), crs=32632)

pts_st = pts[st, ]

# make up some dates
pts_st[["date"]] = sample(seq(as.Date("2000-01-01"), as.Date("2020-01-01"), by="day"), nrow(pts_st))

Now that we have some data, lets query the rainfall

r_data = get_rainfall_point_data(
  pts_st,
  days_back = 10,
  daily_thresh = 1.1,
  n_dry = 2,
  ncores = 2,
  save = T,
  base_path = "~/rainfalldata"
)

But uhhh, there is no column called date_info. So what does this super non-informative message mean?
In the iffi-database (actually only in the already derived product) I put a column that informs about the maximum level of date availbale as only slides with day-information should be processed...
So one (and the only for now;) solution is to add this column in the following way manually:

# as we have all day information in the synthetic dataset...
pts_st[["date_info"]] = "day"

So lets try again to get the rainfall data

r_data = get_rainfall_point_data(
  pts_st,
  days_back = 10,
  daily_thresh = 1.1,
  n_dry = 2,
  ncores = 2,
  save = T,
  base_path = "~/rainfalldata"
)

Ok still does not work....
The problem is a missing unique identifier.
Let's add a unique identifier to each slide

pts_st[["id"]] = 1:nrow(pts_st)

r_data = get_rainfall_point_data(
  pts_st,
  days_back = 10,
  daily_thresh = 1.1,
  n_dry = 2,
  ncores = 2,
  save = T,
  base_path = "~/rainfalldata",
  id_landslide = "id"
)

Ok yet another error. Pfisshhh, what the fluck....
The problem is that the unique identifier should not contain the letters Iand D subsequently in upper or lowercase as this is kind of hardcoded into the PIFF_ID variable...
So lets give them another unique identifier

pts_st[["lll"]] = 1:nrow(pts_st)

r_data = get_rainfall_point_data(
  pts_st,
  days_back = 10,
  daily_thresh = 1.1,
  n_dry = 2,
  ncores = 2,
  save = T,
  base_path = "~/rainfalldata",
  id_landslide = "lll"
)

What, no error...

print(head(r_data))

So we can plot the precpitation for each of these synthetic slide locations

ggplot(r_data) +
  geom_path(
    aes(
      x = days_before_event,
      y = precip,
      group = id
    )
  ) +
  scale_x_reverse() +
  theme_void() +
  labs(
    x = "Days before landslide",
    y = "Daily cumulated precipitation [ mm ]"
  ) +
  theme(
    axis.title = element_text(color="#37323e", size=14, family="Times", face="italic"),
    axis.title.y = element_text(color="#37323e", size=14, family="Times", face="italic", angle=90),
    axis.line.y = element_line(color="#37323e", size=1),
    axis.line.x = element_line(color="#37323e", size=1),
    axis.ticks = element_line(color="black"),
    axis.text = element_text(color="#37323e", family="Times", face="italic", margin = margin(rep(5, 4))),
    plot.margin = unit(rep(2,4), "cm"))