Skip to contents

Process UCDP GED Data with Temporal Fatality Distribution

Source: R/data_ucdp.R
ucdp_ged.Rd

Processes UCDP GED conflict data by distributing fatalities across PRIO-GRID temporal intervals and rasterizing the results. Events spanning multiple time periods have their fatalities proportionally allocated based on temporal overlap, ensuring accurate representation in panel data structures.

Usage

ucdp_ged(
  ged = read_ucdp_ged(),
  violence_types = c(1, 2, 3),
  fatality_variable = "best"
)

Arguments

ged

An sf object containing UCDP GED data. Defaults to read_ucdp_ged() if not provided.

violence_types

Integer vector specifying which types of violence to include. Options are:

  • 1: State-based armed conflict

  • 2: Non-state conflict

  • 3: One-sided violence

Default is c(1, 2, 3) (all types).

fatality_variable

Character string specifying which fatality estimate to use. Options are:

  • "best": Best estimate of fatalities (default)

  • "low": Low estimate of fatalities

  • "high": High estimate of fatalities

  • "event_count": Count of events rather than fatalities

Value

A SpatRaster object with multiple layers, one for each time interval in pg_date_intervals. Layer names correspond to interval end dates. Cell values represent aggregated fatalities (or event counts) for each grid cell and time period.

Details

This function implements a sophisticated fatality distribution algorithm that:

  • Calculates the temporal overlap between each event and PRIO-GRID intervals

  • Distributes fatalities proportionally based on the number of overlapping days

  • Uses integer allocation with remainder distribution to preserve total counts

  • Handles edge cases where events span multiple periods or have single fatalities

  • Validates that distributed fatalities sum to original event totals

The distribution algorithm ensures that:

  • Total fatalities are preserved (no over- or under-counting)

  • Single-fatality events are assigned to exactly one period

  • Fractional allocations are resolved using largest remainder method

Events without sub-national spatial precision (where_prec >= 6) are excluded from the final rasterization, which may result in undercounting in some regions.

Note

  • The function requires the data.table package for efficient processing

  • Processing time scales with the number of events and time intervals

  • A warning is issued if fatality distribution validation fails

  • Events coded at national level or coarser precision are excluded

References

Sundberg R, Melander E (2013). “Introducing the UCDP Georeferenced Event Dataset.” Journal of Peace Research, 50(4), 523–532. ISSN 0022-3433, 1460-3578, doi:10.1177/0022343313484347 , 2024-10-21.

Davies S, Pettersson T, Sollenberg M, Öberg M (2025). “Organized Violence 1989–2024, and the Challenges of Identifying Civilian Victims.” Journal of Peace Research, 62(4), 1223–1240. ISSN 0022-3433, doi:10.1177/00223433251345636 , 2025-06-11.

See also

read_ucdp_ged for loading raw UCDP GED data, ged_ucdp_ged for the standard generator function, pg_date_intervals for PRIO-GRID temporal structure

Examples

if (FALSE) { # \dontrun{
# Process all violence types with best fatality estimates
ged_raster <- ucdp_ged()

# Examine the result
print(ged_raster)

# Process state-based conflict only
state_conflict <- ucdp_ged(violence_types = 1)

# Use high fatality estimates
ged_high <- ucdp_ged(fatality_variable = "high")

# Count events instead of fatalities
event_counts <- ucdp_ged(fatality_variable = "event_count")

# Plot a single time slice
terra::plot(ged_raster[[1]],
            main = paste("UCDP GED Fatalities:", names(ged_raster)[1]))

# Calculate total fatalities across all periods
total_fatalities <- terra::app(ged_raster, sum, na.rm = TRUE)
terra::plot(total_fatalities, main = "Total Fatalities (All Periods)")
} # }