Introduction to imdR

Overview

imdR provides a complete R interface to India Meteorological Department (IMD) gridded daily meteorological data. It covers:

Rainfall — 0.25 degree resolution, 1901 to present
Maximum temperature — 1.0 degree resolution, 1951 to present
Minimum temperature — 1.0 degree resolution, 1951 to present

The package bundles Survey of India (SOI) approved state and district boundaries, enabling boundary-aware extraction and publication-quality maps directly from R.

Two data sources are supported: the quality-controlled archive (get_data(), released yearly) and provisional real-time daily data (get_realtime(), updated with roughly a one-day lag).

Installation

# Install from CRAN (once available)
install.packages("imdR")

# Or install the development version from GitHub
# install.packages("remotes")
remotes::install_github("Subhradip25/imdR")

library(imdR)

Step 1 — Explore available boundaries

# List all 36 states and union territories
list_states()

# List all districts in a state
list_districts("Goa")
list_districts("Kerala")

# Get the sf boundary object for any state or district
goa_sf      <- get_boundary("state",    "Goa")
north_goa   <- get_boundary("district", "North Goa")
kerala_sf   <- get_boundary("state",    "Kerala")

Step 2 — Download archive data

The archive is the authoritative, quality-controlled dataset, released by IMD as full-year files.

file_dir <- tempdir()

# Single year — returns a SpatRaster directly
rain2020 <- get_data("rain", 2020, 2020, file_dir)
tmax2020 <- get_data("tmax", 2020, 2020, file_dir)
tmin2020 <- get_data("tmin", 2020, 2020, file_dir)

# Multi-year — returns a named list of SpatRasters (one per year)
# because leap and non-leap years have different layer counts
rain_3yr <- get_data("rain", 2018, 2020, file_dir)

# Read previously downloaded files without re-downloading
rain2020 <- open_data("rain", 2020, 2020, file_dir)

Step 3 — Download real-time (provisional) data

The archive is released yearly with a lag, so the current year is not available until after it ends. For recent dates — yesterday, this month, or last month — use get_realtime(), which downloads provisional daily grids from IMD’s real-time endpoints.

# Yesterday's rainfall across India
y <- Sys.Date() - 1
r_today <- get_realtime("rain", as.character(y), file_dir = file_dir)

# A date range — e.g. this month to date
first <- format(Sys.Date(), "%Y-%m-01")
r_month <- get_realtime("rain", first, as.character(Sys.Date() - 1),
                        file_dir = file_dir)

# Daily rainfall time series at a location (e.g. Agartala, Tripura)
agartala <- get_realtime("rain", "2026-03-01", "2026-06-23",
                         file_dir = file_dir)
df <- to_csv(agartala, lat = 23.83, lon = 91.28)
head(df)

Important caveats:

Real-time data is provisional and may be revised later by IMD. For final analysis, use get_data() once the archive year is released.
Real-time rainfall is 0.25 degree (same grid as the archive), but real-time temperature is 0.5 degree (61 x 61) — different from the 1.0 degree archive temperature. Do not mix resolutions without resampling.
Variables: "rain", "tmax", "tmin".

Step 4 — Visualize

# Full India map
plot_imd(rain2020, "2020-06-28", "rain")

# Zoom to a state
plot_imd(rain2020, "2020-06-28", "rain",
         level = "state", name = "Kerala",
         save_path = file.path(tempdir(), "rain_Kerala_20200628.png"))

# Zoom to a district
plot_imd(rain2020, "2020-06-28", "rain",
         level = "district", name = "North Goa",
         save_path = file.path(tempdir(), "rain_NorthGoa_20200628.png"))

# Temperature map
plot_imd(tmax2020, "2020-05-20", "tmax",
         save_path = file.path(tempdir(), "tmax_India_20200520.png"))

Step 5 — Point extraction

# Extract daily time series at Panaji, Goa
goa_rain <- get_point(lat = 15.5, lon = 73.8,
                      variable = "rain",
                      start_yr = 2020, end_yr = 2020,
                      file_dir = file_dir)
head(goa_rain)

# Extract all three variables at once (rain + tmax + tmin + DTR)
goa_all <- get_point_all(lat = 15.5, lon = 73.8,
                         start_yr = 2020, end_yr = 2020,
                         file_dir  = file_dir)
head(goa_all)

# Plot the daily time series
plot_timeseries(goa_rain, variable = "rain",
                title = "Goa Daily Rainfall 2020")

Step 6 — District / state time series

get_district_series() returns a tidy daily time series of the spatial mean across all grid cells in a state or district — the correct way to summarise an area (average over space). To get a seasonal or annual total for rainfall, sum the returned daily values (rainfall accumulates over time).

# District rainfall — multi-year daily series, saved as CSV
karnal <- get_district_series("rain", "Karnal",
                              start = 2018, end = 2020,
                              file_dir = file_dir)
head(karnal)

# Annual total = sum of daily means
sum(karnal$rain, na.rm = TRUE)

# State temperature with a publication-quality plot.
# Goa is smaller than a 1.0 degree temperature cell, so touches = TRUE
# captures the overlapping cells.
get_district_series("tmax", "Goa",
                    start = 2020, end = 2020,
                    level = "state",
                    file_dir = file_dir,
                    touches = TRUE,
                    plot = TRUE)

# Real-time district series — recent provisional data
get_district_series("rain", "North Tripura",
                    start = "2026-06-01", end = "2026-06-25",
                    file_dir = file_dir,
                    realtime = TRUE, touches = TRUE)

Step 7 — Bounding box and boundary extraction

# Crop to a bounding box
western_ghats <- get_bbox(
  lat_min = 8, lat_max = 21,
  lon_min = 73, lon_max = 78,
  variable = "rain", start_yr = 2020, end_yr = 2020,
  file_dir = file_dir, format = "netcdf"
)

# Mask to a state boundary
kerala_rain <- extract_by_boundary(
  rain2020, level = "state", name = "Kerala",
  variable = "rain", save = TRUE,
  format = "netcdf", file_dir = file_dir
)

# Mask to a district boundary
north_goa_rain <- extract_by_boundary(
  rain2020, level = "district", name = "North Goa",
  variable = "rain", save = TRUE,
  format = "geotiff", file_dir = file_dir
)

Step 8 — Rainfall climate indices

# Compute 11 indices for full India (2020)
rain_idx <- compute_rainfall_indices(rain2020,
                                     file_dir = file_dir)

# Available indices:
# dr     — rainy days (>= 2.5 mm)
# d64    — heavy precipitation days (>= 64.5 mm)
# d115   — very heavy precipitation days (>= 115.6 mm)
# rx1day — maximum 1-day rainfall
# rx5day — maximum 5-day rainfall
# rtwd   — total rainfall on wet days
# sdii   — simple daily intensity index
# total  — annual total rainfall
# cwd    — consecutive wet days
# cdd    — consecutive dry days
# pci    — precipitation concentration index

# Compute for Goa state only
goa_rain_idx <- compute_rainfall_indices(
  rain2020, level = "state", name = "Goa",
  file_dir = file_dir
)
print(goa_rain_idx)

# Multi-year indices for trend analysis
idx_3yr <- compute_rainfall_indices(rain_3yr,
                                    file_dir = file_dir)

Step 9 — Temperature climate indices

# Compute 13 temperature indices for full India
temp_idx <- compute_temp_indices(tmax2020, tmin2020,
                                  file_dir = file_dir)

# Available indices:
# mean_tmax — mean daily maximum temperature
# mean_tmin — mean daily minimum temperature
# mean_dtr  — mean diurnal temperature range
# txx       — hottest day (max of tmax)
# txn       — coldest day (min of tmax)
# tnx       — warmest night (max of tmin)
# tnn       — coldest night (min of tmin)
# su35      — summer days (tmax >= 35 C)
# su40      — very hot days (tmax >= 40 C)
# tr10      — cold nights (tmin <= 10 C)
# tr25      — tropical nights (tmin >= 25 C)
# wsdi      — warm spell duration index
# csdi      — cold spell duration index

# Compute for Goa
goa_temp_idx <- compute_temp_indices(
  tmax2020, tmin2020,
  level = "state", name = "Goa",
  file_dir = file_dir
)
print(goa_temp_idx)

Step 10 — Trend analysis

# Mann-Kendall test + Sen's slope on any index
# Requires at least 3 years of data

# Download a longer time series for meaningful trend
rain_10yr <- get_data("rain", 2011, 2020, file_dir)
idx_10yr  <- compute_rainfall_indices(rain_10yr,
                                       file_dir = file_dir)

# Trend in annual total rainfall
trend_total <- trend_analysis(idx_10yr,
                              index_col = "total",
                              file_dir  = file_dir)

# Trend in rainy days
trend_dr <- trend_analysis(idx_10yr,
                           index_col = "dr",
                           file_dir  = file_dir)

# Region-specific trend — Goa
goa_idx_10yr <- compute_rainfall_indices(
  rain_10yr, level = "state", name = "Goa",
  file_dir = file_dir
)
trend_goa <- trend_analysis(goa_idx_10yr,
                            index_col = "total",
                            name      = "Goa",
                            file_dir  = file_dir)

Output files

All functions that produce files use consistent naming conventions:

Function	Output file
`get_point()`	`rain_15.5N_73.8E_2020_2020.csv`
`get_point_all()`	`imd_all_15.5N_73.8E_2020_2020.csv`
`get_district_series()`	`rain_Karnal_series_2018_2020.csv`
`get_bbox()`	`rain_8N_21N_73E_78E_2020_2020.nc`
`extract_by_boundary()`	`rain_Kerala_2020-01-01_2020-12-31.nc`
`compute_rainfall_indices()`	`rainfall_indices_Goa.csv`
`compute_temp_indices()`	`temp_indices_Goa.csv`
`trend_analysis()`	`trend_total_Goa.csv` + `trend_total_Goa.png`