scripts/tif2csv.R

{
  setwd("~/Downloads")
  
  # Download the GeoTIFF file
  url <- "https://huggingface.co/datasets/cjerzak/TemporalNeighborhoodMaterialWealthAfrica/resolve/main/wealth_map.tif"
  download.file(url, "wealth_map.tif", mode = "wb")
  
  # Load the terra package
  library(terra)
  
  # Read the multiband raster
  rast <- rast("wealth_map.tif")
  
  # Assign names to the bands based on time periods
  periods <- c("1990-1992", "1993-1995", "1996-1998", "1999-2001", "2002-2004", "2005-2007", "2008-2010", "2011-2013", "2014-2016", "2017-2019")
  names(rast) <- periods
  
  # Convert to data frame with longitude and latitude, removing rows with NA values
  df <- as.data.frame(rast, xy = TRUE, na.rm = TRUE)
  df <- df[rowMeans(df[,-c(1:2)])!=0,]
  
  # Rename coordinate columns
  names(df)[1:2] <- c("longitude", "latitude")
  
  # Write to CSV
  write.csv(df, "wealth_africa.csv", row.names = FALSE)
  
  # Write to paraquet 
  arrow::write_parquet(df, "wealth_africa.parquet")
}