Nighttime lights have become a commonly used resource to estimate changes in local economic activity. This section shows where nighttime lights are concentrated across Syria and observed changes over time.
Data
We use nighttime lights data from the VIIRS Black Marble dataset. Raw nighttime lights data requires correction due to cloud cover and stray light, such as lunar light. The Black Marble dataset applies advanced algorithms to correct raw nighttime light values and calibrate data so that trends in lights over time can be meaningfully analyzed. From VIIRS Black Marble, we use data from January 2012 through present—where data is available at a 500-meter resolution.
The following code downloads and processes data: _main.R Defines filepaths and loads packages 01_download_ntl.R Downloads NTL data 02_extract_data.R Extracts/aggregates nighttime light data to different polygons
Map of nighttime lights
We first show a map of nighttime lights. Most of the country is dark, with lights concentrated within cities.
Code
## Load boundariesadm0_sf <-read_sf(file.path(boundaries_dir, "ner_adm_ignn_20230720_ab_shp","NER_admbnda_adm0_IGNN_20230720.shp"))## Load/prep rasterprep_r <-function(year_i){ r <-rast(file.path(ntl_dir, "individual_rasters", "annually",paste0("VNP46A4_NearNadir_Composite_Snow_Free_qflag_t",year_i,".tif"))) r <- r %>%mask(adm0_sf) r[][r[] ==0] <-NA r[] <-log(r[] +1) r[] <-log(r[] +1)return(r)}r_2012 <-prep_r(2012)r_2013 <-prep_r(2013)r_2014 <-prep_r(2014)r_2015 <-prep_r(2015)r_2016 <-prep_r(2016)r_2017 <-prep_r(2017)r_2018 <-prep_r(2018)r_2019 <-prep_r(2019)r_2020 <-prep_r(2020)r_2021 <-prep_r(2021)r_2022 <-prep_r(2022)r_2023 <-prep_r(2023)## Make mappal <-colorNumeric(c("yellow", "orange", "red"), unique(c(r_2012[], r_2013[], r_2014[], r_2015[], r_2016[], r_2017[], r_2018[], r_2019[], r_2020[], r_2021[], r_2022[], r_2023[])),na.color ="transparent")leaflet() %>%addProviderTiles(providers$CartoDB.DarkMatter) %>%addRasterImage(r_2012, colors = pal, opacity =1, group ="2012") %>%addRasterImage(r_2013, colors = pal, opacity =1, group ="2013") %>%addRasterImage(r_2014, colors = pal, opacity =1, group ="2014") %>%addRasterImage(r_2015, colors = pal, opacity =1, group ="2015") %>%addRasterImage(r_2016, colors = pal, opacity =1, group ="2016") %>%addRasterImage(r_2017, colors = pal, opacity =1, group ="2017") %>%addRasterImage(r_2018, colors = pal, opacity =1, group ="2018") %>%addRasterImage(r_2019, colors = pal, opacity =1, group ="2019") %>%addRasterImage(r_2020, colors = pal, opacity =1, group ="2020") %>%addRasterImage(r_2021, colors = pal, opacity =1, group ="2021") %>%addRasterImage(r_2022, colors = pal, opacity =1, group ="2022") %>%addRasterImage(r_2023, colors = pal, opacity =1, group ="2023") %>%addLayersControl(baseGroups =paste0(2012:2023),options =layersControlOptions(collapsed=FALSE) )
Association of NTL with GDP
Existing research demonstrates that nighttime lights correlates with GDP, serving as a proxy for economic activity (for example, see here). Below we show the association between nighttime lights and GDP for Niger at the annual and country level.
The below section shows the percent change in nighttime lights from August to December 2023 relative to August to December 2022. We show the percent change at the pixel level and at the ADM3 level.