Skip to contents

Process raster data with administrative boundaries

Source: R/process_rasters.R
process_raster_with_boundaries.Rd

This function processes a single raster file by extracting zonal statistics within administrative boundaries. It handles CHIRPS rainfall data format by default, extracting dates from filenames, but works with any raster data.

Usage

process_raster_with_boundaries(
  raster_file,
  shapefile,
  id_cols = c("adm0", "adm1"),
  aggregations = c("mean"),
  raster_is_density = FALSE,
  layer_to_process = 1,
  pop_raster_file = NULL,
  weight_na_as_zero = TRUE,
  pattern_info = NULL,
  time_components = NULL
)

Arguments

raster_file

Path to raster file (.tif format)

shapefile

sf/SpatVector object with administrative boundaries. Must be in the same coordinate reference system (CRS) as the raster data

id_cols

Vector of column names from the shapefile to include in output. Default is c("adm0", "adm1") for country and first admin level

aggregations

Vector of aggregation methods to calculate. Supported values are "mean", "sum", "median", "weighted_mean", and "weighted_median". The two weighted options require pop_raster_file. Default is c("mean")

raster_is_density

Logical indicating if raster values represent density. If TRUE, values are converted from density to counts using cell area. Default is FALSE.

layer_to_process

Integer or character. Specifies which layer in a multi-layer raster to extract. If the raster contains multiple layers (e.g., different years or indicators), this argument selects the layer to be processed. Default is 1

pop_raster_file

Optional path to a population raster used as weights for "weighted_mean" and "weighted_median" aggregations. Reprojected and resampled to match raster_file. Default is NULL.

weight_na_as_zero

Logical. If TRUE, missing weights are treated as zero; if FALSE, as NA. Only used when weighted aggregations are requested. Default is TRUE.

pattern_info

Optional. Pre-parsed time pattern from detect_time_pattern(). If NULL, detected from filename.

time_components

Optional. Pre-parsed time components from extract_time_components(). If NULL, extracted from filename.

Value

A data frame containing:

  • file_name: Name of processed raster file

  • Selected ID columns from shapefile (specified by id_cols)

  • Time columns based on detected pattern: - For monthly data: year, month - For yearly data: year

  • Aggregated values with column names matching aggregation methods

Details

The function performs the following steps:

  1. Validates requested aggregation methods

  2. Detects and extracts date information from filename

  3. Loads raster data and handles no-data values (-9999)

  4. If raster_is_density is TRUE, converts density to counts using cell area

  5. Calculates zonal statistics using exactextractr. When weighted aggregations are requested, loads the population raster, reprojects and resamples it to match the value raster, then uses it as per-cell weights

  6. Combines results with shapefile attributes

  7. Returns a clean data frame without geometry

Examples

if (FALSE) { # \dontrun{
# Process single CHIRPS file with district-level boundaries
districts <- sf::st_read("districts.shp")
results <- process_raster_with_boundaries(
  raster_file = "chirps-v2.0.2022.01.tif",
  shapefile = districts,
  id_cols = c("COUNTRY", "DISTRICT"),
  aggregations = c("mean", "sum", "median")
)

# Population-weighted statistics alongside plain ones
results_weighted <- process_raster_with_boundaries(
  raster_file = "pfpr_2022.tif",
  shapefile = districts,
  id_cols = c("COUNTRY", "DISTRICT"),
  aggregations = c("mean", "weighted_mean", "weighted_median"),
  pop_raster_file = "worldpop_2022.tif"
)
} # }