Climate data for Brazilian municipalities

Raphael Saldanha et al.

Fiocruz

2024-11-20

Overview

  • Climate raw data sources: ground level and reanalysis
  • Data on Braziliam municipalities spatial unit
  • Climatological normals and indicators

Ground-level climate data

  • Weather stations on Cametá and Mocajuba (north region)
  • Data every 5 minutes, sent to Fiocruz Postgres server
  • plugfieldapi package to retrieve data from stations with daily e-mail reports
  • Error detection and notification for problems

ReAnalysis climate data

  • ERA5-Land: global coverage, hourly and daily data, regular updates
  • BR-DWGD: Brazil coverage, daily data, sensible to extreme events, sporadic updates
  • TerraClimate: global coverage, monthly data, higher resolution (~4km), regular updates

More details here

Climate data for Brazilian municipalities

  • Zonal statistics computation
  • Adoption of exactextractr package for cell’s coverage weighted computations
  • Creation of package zonalclim with helper functions to compute scalable zonal statistics with chunks strategy
  • DAG system using the targets package to compute climate zonal statistics for Brazilian municipalities
  • Publication on Environmental Data Science journal

More details here

Example

Climatological normals

  • Climatological normals are computed only for weather stations and municipal references are needed for better climate change understanding
  • Normals computed for each Brazilian municipality using the Zonal BR-DWGD from 1961 to 1990
    • Mean, 10th and 90th percentile
    • Temperature (max, min), precipitation, relative humidity, solar radiation, wind speed, and evapotranspiration

Time-aggregated indicators

  • Municipal daily data time series to monthly indicators
  • Monthly statistics
    • Average, median, standard deviation, standard error, maximum and minimum values, and percentiles
  • Occurence of events
    • 22 indicators, in reference to normals or count sequence. Creation of nseq and climindi packages
    • Heat waves, cold spells, count of warm days, count of dry and wet days, and others

More details here.

Future work

  • Dashboards for weather stations
  • Compute zonal statistics with populated areas weights
  • Create other time-aggregated indicators
  • Test the methodologies on other countries
  • Compare ground-level data from weather stations with climate reanalysis datasets