Climate and Environmental Covariate Extraction for DHS Clusters in Sierra Leone

Read full text Summarise with VIRAT
Authors: Ishmael Sandy
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Understood. Here is the text you can paste directly into Zenodo's description field. Climate and Environmental Covariate Extraction for DHS Clusters in Sierra Leone Description Overview This repository contains a Google Earth Engine (GEE) script that extracts spatially explicit climate, environmental, and infrastructure covariates at Demographic and Health Survey (DHS) cluster locations in Sierra Leone. The script processes the DHS survey waves for 2008, 2013, and 2019, and produces a harmonised cluster-level CSV dataset intended for econometric analysis of climate impacts on household welfare. Version 10 is a substantially revised release. Key changes from prior versions include: correction of the differential baseline end year (previously set to SURVEY_YEAR minus 3; corrected to SURVEY_YEAR minus 1); resampling of CHIRPS daily precipitation images to 3 km before annual totals are computed; and the addition of a fixed 1981 to 2005 baseline that applies uniformly across all three survey waves to support identification strategies requiring strict exogeneity. The script was developed with the assistance of artificial intelligence tools and subsequently reviewed, modified, and fully supervised by the author. All methodological decisions, including variable definitions, baseline construction, anomaly thresholds, wind direction verification, scaling procedures, and resampling rules, were designed and validated by the author. Survey Waves The script processes three DHS survey waves for Sierra Leone: 2008 wave: asset SLGE53FL_2008, differential baseline end year 2007 2013 wave: asset SLGE61FL_2013, differential baseline end year 2012 2019 wave: asset SLGE7AFL_2019, differential baseline end year 2018 Cluster coordinates are loaded from GEE assets stored under projects/ee-ishmaelsandy77/assets/. Each cluster is assigned administrative names at levels 1 and 2 using FAO GAUL 2015 boundaries. Spatial Aggregation Design Climate and environmental variables are extracted using a median reducer within a 10 km buffer surrounding each DHS cluster centroid. This buffer size reflects the maximum random spatial displacement applied to DHS cluster coordinates to protect respondent confidentiality. Distance and cost variables are extracted within a smaller 5 km buffer to reduce aggregation bias near linear infrastructure features. All extraction uses a uniform 10 km scale (EXTRACTION_SCALE = 10,000 m), which matches the native resolution of ERA5-Land reanalysis data (approximately 0.1 degree, roughly 11 km at 8.5 degrees North). CHIRPS daily images (native resolution approximately 5.5 km) are resampled to 3 km using bilinear interpolation before annual totals are computed. The 3 km target equals one-third of the buffer distance, ensuring adequate pixel sampling density within each 10 km buffer during zonal statistics extraction. Baseline Design Two precipitation baseline windows are constructed. The fixed baseline runs from 1981 to 2005 and applies identically across all three DHS waves. Because 2005 predates all three survey years, this baseline is strictly exogenous and constitutes the main specification. The differential baseline runs from 1981 to SURVEY_YEAR minus 1 and is provided for robustness checks. A three-year pre-survey average (SURVEY_YEAR minus 3 to SURVEY_YEAR minus 1) is also constructed; this reduces anomaly variance compression in the 2008 wave and produces comparable standard deviations across waves. For each baseline, the script computes the long-run median, mean, and standard deviation of annual totals, which serve as reference values for anomaly calculations. Climate Variables CHIRPS precipitation (source: UCSB-CHG/CHIRPS/DAILY, resampled to 3 km before use): chirps_current_mm: annual precipitation total for the survey year chirps_fixbase_median_mm, chirps_fixbase_mean_mm, chirps_fixbase_stddev_mm: fixed baseline (1981 to 2005) statistics chirps_abs_anom_fixmedian_mm, chirps_std_anom_fixmedian: absolute and standardised anomaly from fixed baseline median chirps_abs_anom_fixmean_mm, chirps_std_anom_fixmean: absolute and standardised anomaly from fixed baseline mean chirps_diffbase_median_mm, chirps_diffbase_mean_mm, chirps_diffbase_stddev_mm: differential baseline statistics chirps_abs_anom_diffmedian_mm, chirps_std_anom_diffmedian: anomalies from differential baseline median chirps_abs_anom_diffmean_mm, chirps_std_anom_diffmean: anomalies from differential baseline mean chirps_3yr_avg_mm: mean of annual totals for SURVEY_YEAR minus 3 to SURVEY_YEAR minus 1 chirps_abs_anom_3yr_fix_mm, chirps_std_anom_3yr_fix: three-year anomaly against fixed baseline (recommended main precipitation regressor) chirps_3yr_window: start year of the three-year window ERA5-Land precipitation (source: ECMWF/ERA5_LAND/MONTHLY_AGGR, cross-validation): same 15-variable structure as CHIRPS with prefix era5p_, plus era5p_3yr_avg_mm, era5p_abs_anom_3yr_fix_mm, era5p_std_anom_3yr_fix, era5p_3yr_window. ERA5-Land wind and surface pressure: era5_u_wind_10m, era5_v_wind_10m: annual mean eastward and northward wind components (m/s) era5_wind_speed: vector wind speed (m/s) era5_wind_dir_deg: meteorological FROM-direction, 0 to 360 degrees clockwise from North era5_southerly_flow: binary indicator, northward component greater than 0 (West African Monsoon inflow) era5_surfpres_current_pa: annual mean surface pressure for the survey year (Pa) era5_surfpres_fixbase_median_pa, era5_surfpres_fixbase_mean_pa: fixed baseline surface pressure statistics era5_surfpres_diffbase_median_pa, era5_surfpres_diffbase_mean_pa: differential baseline surface pressure statistics Temperature (source: ERA5-Land band temperature_2m, converted from Kelvin to degrees Celsius): temp_current_c: annual mean 2 m air temperature for the survey year temp_fixbase_median_c, temp_fixbase_mean_c: fixed baseline statistics temp_anom_fixmedian_c, temp_anom_fixmean_c: temperature anomalies from fixed baseline temp_diffbase_median_c, temp_diffbase_mean_c: differential baseline statistics temp_anom_diffmedian_c, temp_anom_diffmean_c: temperature anomalies from differential baseline Environmental and Infrastructure Variables Particulate matter (source: Global Satellite-Derived PM2.5 Version 6.GL.02, Shen et al., 2024, DOI: 10.1021/acsestair.3c00054; coverage 2000 to 2022; both baselines truncated to start at 2000): pm25_current_ug_m3: annual mean PM2.5 for the survey year (micrograms per cubic metre) pm25_fixbase_median_ug_m3, pm25_fixbase_mean_ug_m3: fixed baseline (2000 to 2005) statistics pm25_diffbase_median_ug_m3, pm25_diffbase_mean_ug_m3: differential baseline statistics pm25_fixbase_start, pm25_diffbase_start: baseline start year documentation (both equal 2000) Terrain (source: SRTM 30 m, USGS/SRTMGL1_003): elevation: mean elevation (metres) slope: mean terrain slope (degrees) aspect_degrees: mean terrain aspect (degrees) terrain_ruggedness_index: standard deviation of elevation within a 1-pixel neighbourhood Soil quality (source: SoilGrids ISRIC, 0 to 5 cm depth layer): soil_quality_index: composite index calculated as SOC/(SOC+1) + clay/(clay+100) minus |pH minus 7|/7 Nighttime lights: nightlights_1992: DMSP-OLS stable_lights mean, 1992 (source: NOAA/DMSP-OLS/NIGHTTIME_LIGHTS) nightlights_survey_ref_year: DMSP-OLS stable_lights for the year preceding the survey (2008 and 2013 waves); VIIRS avg_rad for 2018 (2019 wave). Note: DMSP and VIIRS values are not directly comparable. Flood risk (source: JRC Global Surface Water version 1.4): flood_occurrence_pct: percentage of years with surface water presence flood_recurrence_pct: percentage of wet years relative to all observed years flood_seasonality_months: number of months per year with surface water flood_prone: binary indicator, occurrence greater than 0 and less than 100 percent Road infrastructure and distance variables (extracted at 5 km point buffer): distance_coast_km: Euclidean distance to nearest coastline (source: NOAA ETOPO1) distance_paved_road_km: least-cost distance to nearest paved road (source: GRIP4 Africa, GP_RSE = 1) distance_unpaved_road_km: least-cost distance to nearest unpaved road (source: GRIP4 Africa, GP_RSE = 2) paved_road_share: share of paved roads among all classified roads within 10 km dist_colonial_hq_km: Euclidean distance to nearest of 9 colonial district capitals (kilometres) hist_lcp_port_min: terrain-weighted travel time to nearest colonial port (Freetown, Bonthe, Sulima) using Tobler's hiking function (minutes) hist_lcp_hq_min: terrain-weighted travel time to nearest colonial administrative headquarters (minutes) Administrative variables (source: FAO GAUL 2015): adm1_name, adm2_name: province and district names area_sqkm: district area in square kilometres Metadata columns DHSCLUST, LATNUM, LONGNUM: cluster identifier and coordinates country: country code (SL) survey_year: DHS survey year baseline_start: 1981 for all waves fixbase_end: 2005 for all waves diffbase_end: SURVEY_YEAR minus 1 precip_resample_scale: 3000 (documents the 3 km CHIRPS resampling) extraction_scale: 10000 (documents the 10 km zonal statistics scale) Data Sources CHIRPS Daily: UCSB-CHG/CHIRPS/DAILY ERA5-Land Monthly: ECMWF/ERA5_LAND/MONTHLY_AGGR SRTM 30 m: USGS/SRTMGL1_003 SoilGrids ISRIC: projects/soilgrids-isric/ DMSP-OLS Nighttime Lights: NOAA/DMSP-OLS/NIGHTTIME_LIGHTS VIIRS Monthly: NOAA/VIIRS/DNB/MONTHLY_V1/VCMCFG JRC Global Surface Water 1.4: JRC/GSW1_4/GlobalSurfaceWater GRIP4 Africa Roads: projects/sat-io/open-datasets/GRIP4/Africa NOAA ETOPO1: NOAA/NGDC/ETOPO1 Global Satellite PM2.5 V6.GL.02: projects/sat-io/open-datasets/GLOBAL-SATELLITE-PM25/ANNUAL FAO GAUL 2015: FAO/GAUL/2015/level1 and level2 Reference for PM2.5 data: Shen, S., Li, C., van Donkelaar, A., Jacobs, N., Wang, C., and Martin, R. V. (2024). Enhancing global estimation of fine particulate matter concentrations by

Fields of Study & Subjects

Climate change impacts on agriculture Climate Change and Health Impacts Climate variability and models

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
25 March 2026
Language
EN