README

climateBR was developed to help social scientists operationalize research on climate shocks in Brazil. The package provides tools to download, process, and analyze historical meteorological data from the Brazilian National Institute of Meteorology (INMET).

INMET station data are available from 2000 to the present. However, I recommend focusing analyses on data from 2008 onward, as there are substantially more observations available from this period. Before 2008, fewer than 200 stations were operating.

Installation

The package is available on CRAN and can be installed using the standard R installation method:

install.packages("climateBR")

If you want to test the latest development version, which may include new functions and features, you can install it from GitHub:

# install.packages("remotes")

remotes::install_github("kaiorb52/climateBR")

Examples

With climateBR, you can download historical INMET data using the download_inmet() function, specifying one or more years.

The downloaded raw station CSV files can then be processed into a partitioned Apache Arrow dataset using build_inmet_dataset(). Finally, read_inmet() can be used to query the processed dataset.

library(climateBR)

download_inmet(
  years = 2026,
  unzip_to = "data/raw/inmet"
)

build_inmet_dataset(
  input = "data/raw/inmet/",
  output = "data/processed/inmet/"
)

df_inmet <- read_inmet(path = "data/processed/inmet")

The package also provides datasets containing information on the nearest INMET stations to each municipality in Brazil.

data("mun_stations")
# The `mun_stations` dataset was created using the `nearest_stations()` function and contains the five nearest stations for each municipality.

temp_rj <- df_inmet |> 
  group_by(ano, mes, codigo_wmo) |> 
  summarise(
    temp_max = max(
      temperatura_maxima_na_hora_ant_aut_c,
      na.rm = TRUE
    )
  ) |> 
  left_join(
    mun_stations |> 
      filter(station_order == 1) |> 
      select(state_muni, code_ibge7, code_wmo), 
    by = c("codigo_wmo" = "code_wmo")
  ) |> 
  filter(state_muni == "RJ") |> 
  collect() |> 
  arrange(-temp_max)
  
# A tibble: 644 × 6
# # Groups:   ano, mes [7]
#      ano   mes codigo_wmo temp_max state_muni code_ibge7
#    <int> <int> <chr>         <dbl> <chr>           <dbl>
#  1  2026     1 A601           41   RJ            3305554
#  2  2026     1 A601           41   RJ            3304144
#  3  2026     1 A601           41   RJ            3303609
#  4  2026     1 A601           41   RJ            3302270
#  5  2026     1 A601           41   RJ            3302007
#  6  2026     1 A621           40.8 RJ            3305109
#  7  2026     1 A621           40.8 RJ            3303500
#  8  2026     1 A621           40.8 RJ            3303203
#  9  2026     1 A621           40.8 RJ            3302858
# 10  2026     1 A621           40.8 RJ            3300456
# # ℹ 634 more rows
# # ℹ Use `print(n = ...)` to see more rows


library(ggplot)

temp_rj |> 
  ggplot(aes(x = as.character(mes), y = temp_max)) +
  geom_boxplot() +
  theme_linedraw() +
  labs(y = "Max. Temp (Cº)", x = "Month") +
  ylim(20, 40)


library(geobr)

mun_24 <- geobr::read_municipality(year = 2024)

mun_rj <- mun_24 |> 
  filter(abbrev_state == "RJ") |> 
  select(code_muni, geom)

mun_temp_rj <- mun_rj |> 
  left_join(
    mun_stations |> filter(station_order == 1) |> select(code_ibge7, code_wmo), 
    by = c("code_muni" = "code_ibge7")
  ) |> 
  left_join(
    temp_rj, 
    by = c("code_wmo" = "codigo_wmo")
  )

mun_temp_rj$mes <- factor(
  mun_temp_rj$mes,
  levels = 1:12,
  labels = month.name
)

ggplot() +
  geom_sf(data = mun_temp_rj, aes(fill = temp_max)) +
  scale_fill_distiller(palette = "RdYlGn") +
  facet_wrap(.~mes) +
  labs(fill = "Max. Temp (Cº)") +
  theme_void() +
  theme(
    legend.position = c(0.785, 0.185),
    plot.background = element_rect(fill = "white")
  )

climateBR

Installation

Examples

License

Citation