Ingest GeoTIFFs as Predictors • evoland

In a likely analytic scenario, your analytic workflow will include the ingestion of predictor data from disk instead of a remote source. This is not ideal for reproducibility - in the best case, you would be retrieving data from a persistent archive that uses a DOI handle.

This file shows how to work in either scenario. Let’s start by attaching a database to work in, and setting some coordinate points

library(evoland)
library(data.table)
db <- evoland_db$new(path = "noisemodel.evolanddb")
db$set_coords(
  type = "square",
  epsg = 2056,
  extent = terra::ext(c(
    xmin = 2697000,
    xmax = 2698000,
    ymin = 1252000,
    ymax = 1253000
  )),
  resolution = 100
)

db$set_periods(
  period_length_str = "P10Y",
  start_observed = "1985-01-01",
  end_observed = "2020-01-01",
  end_extrapolated = "2060-01-01"
)

Ingest Predictors from Online Sources

Here, we’ll start off with a relatively simple predictor: The current level of noise pollution, in accordance with the current levels of traffic.

coords_minimal <- db$coords_minimal

sonbase_spec <- list(
  noise = list(
    unit = "dBa",
    pretty_name = "Maximum noise exposure",
    orig_format = "10m*10m raster",
    description = "The maximum across all sonBASE noise exposure categories, i.e. daytime & nighttime road & rail exposure",
    sources = list(
      list(
        url = "https://data.geo.admin.ch/ch.bafu.laerm-strassenlaerm_tag/laerm-strassenlaerm_tag/laerm-strassenlaerm_tag_2056.tif",
        md5sum = "09791808dbf12fcde82182e70f2ebdfb"
      ),
      list(
        url = "https://data.geo.admin.ch/ch.bafu.laerm-strassenlaerm_nacht/laerm-strassenlaerm_nacht/laerm-strassenlaerm_nacht_2056.tif",
        md5sum = "6e79dc1d353751084e21dc6b61778b99"
      )
      # list(
      #   url = "https://data.geo.admin.ch/ch.bafu.laerm-bahnlaerm_nacht/laerm-bahnlaerm_nacht/laerm-bahnlaerm_nacht_2056.tif",
      #   md5sum = "161df62f9a2a29c9120380f965aa19ba"
      # ),
      # list(
      #   url = "https://data.geo.admin.ch/ch.bafu.laerm-bahnlaerm_tag/laerm-bahnlaerm_tag/laerm-bahnlaerm_tag_2056.tif",
      #   md5sum = "6016b9ca4c974cb982fbe18112f201fe"
      # )
    )
  )
)

sonbase_sources <-
  sonbase_spec$noise$sources |>
  data.table::rbindlist() |>
  download_and_verify()

# data is at 10m
# extent plus 1km, given metres for unit; enough for all resampling strategies
extent_wide <- db$extent |> terra::extend(1000)

sonbase_max <-
  purrr::map(
    sonbase_sources$local_path,
    \(x) terra::rast(x) |> terra::crop(extent_wide)
  ) |>
  terra::rast() |>
  max(na.rm = TRUE) |>
  terra::resample(10, method = "bilinear") |> # downsample factor 10: to hectare
  extract_using_coords_t(coords_minimal)

db$add_predictor(
  pred_spec = sonbase_spec,
  pred_data = sonbase_max[
    ,
    # the id_period = 0L is a 0 integer; this special period indicates "static" data
    # that's not associated with a historical period
    .(id_coord, id_period = 0L, value)
  ],
  pred_type = "float"
)

Check Results

Let’s see how the data looks in the DB.

db$pred_meta_t

We can see that the noise predictor was assigned id_pred = 1, which makes sense, given that this was the first predictor to be ingested

db$pred_data_t_float

We see that there’s currently a single predictor in our table at 100 coordinate points. As expected, once the metadata was inserted into the database, we used the allotted identifier to label our inserted sonbase_max data.