Skip to contents

Get Weather Data for Disease Modelling

Source: R/get_wth.R
get_wth.Rd

Retrieves daily weather data from either the NASA POWER database or ERA5 reanalysis (via the Open-Meteo Archive API) and formats it for use in seir() and epicrop disease models.

Usage

get_wth(lonlat, dates, duration = NULL, source = c("power", "era5"))

Arguments

lonlat

Numeric vector of coordinates as c(longitude, latitude). Longitude: -180 to 180 degrees (West/East) Latitude: -90 to 90 degrees (South/North) Example: c(121.255669, 14.167425) for IRRI, Philippines

dates

Character vector of start date and optional end date, or character scalar start date. Format: "YYYY-MM-DD" (ISO 8601).

  • Length 1: Use with duration parameter

  • Length 2+: First is start, second is end (used only if duration is not supplied)

duration

Optional integer. Number of days to retrieve (inclusive). If supplied, it takes precedence over any end date in dates. Must be positive (> 0). If not provided and dates has only one element, an error is raised.

source

Character. Weather data source, one of "power" (default) or "era5".

Value

An epicrop.wth object (data.table::data.table() subclass) containing daily weather data with columns:

ColumnDescriptionUnits
YYYYMMDDDateISO 8601 (YYYY-MM-DD)
DOYDay of yearInteger (1-365/366)
TEMPMean daily temperatureDegrees C
TMINMinimum daily temperatureDegrees C
TMAXMaximum daily temperatureDegrees C
RHUMMean daily relative humidityPercent (0-100)
RAINTotal daily rainfallMillimeters
LATLatitude of query locationDegrees (-90 to 90)
LONLongitude of query locationDegrees (-180 to 180)

Ready for direct use in seir(), bacterial_blight(), leaf_blast(), and other disease models.

Data Source

  • "power" (default): NASA POWER via nasapower::get_power(), global daily data from 1984 onwards at 0.5-degree resolution.

  • "era5": ERA5 from 1940 onwards at about 0.25-degree resolution, roughly 25 km, via the Open-Meteo Archive API (https://open-meteo.com/en/docs/historical-weather-api). Hourly values are aggregated to daily values (mean/min/max temperature, mean relative humidity, total precipitation) by local calendar day so that the output matches the POWER-derived columns. ERA5 data are typically available with a lag of about five days from 1940 onwards.

Both sources are modelled/reanalysis products and may differ from local weather station observations. Note that ERA5 precipitation is not the same quantity as POWER's bias-corrected PRECTOTCORR, so rainfall totals will differ between sources.

Citing Weather Data

Please be kind and cite the data and their sources. To do so, please see the NASA POWER and Openmeteo websites for how to cite the data when you use them.

Retrieved Variables

This function automatically provides:

  • Mean daily temperature (2-meter height)

  • Daily minimum temperature

  • Daily maximum temperature

  • Mean daily relative humidity (2-meter height)

  • Total daily precipitation (bias-corrected for POWER)

All variables are provided in standard units: temperature in degrees C, humidity in %, rainfall in mm.

Date Handling

Two methods for specifying the weather window:

  1. Start and end dates (character vector): dates = c("2000-06-30", "2000-12-31")

  2. Start date + duration (integer): dates = "2000-06-30", duration = 120

Duration is interpreted as inclusive days: end = start + duration - 1. For example, duration = 120 starting on June 30 ends on October 27 (120 days total).

References

NASA POWER project (2019). Agroclimatology, Prediction Of Worldwide Energy Resource. Funded by NASA's Applied Science Program. https://power.larc.nasa.gov/.

Hersbach, H. et al. (2020). The ERA5 global reanalysis. Quarterly Journal of the Royal Meteorological Society, 146, 1999-2049. doi:10.1002/qj.3803 .

Sparks, A. H., P. D. Esker, M. Bates, W. Dall'Acqua, Z. Guo, V. Segovia, S. D. Silwal, S. Tolos, and K. A. Garrett. (2008). Ecology and Epidemiology in R: Disease Progress over Time. The Plant Health Instructor. doi:10.1094/PHI-A-2008-0129-02 .

Zippenfenig, P. (2023) Open-Meteo.com Weather API. Zenodo. doi: doi:10.5281/ZENODO.7970649 .

Author

Adam H. Sparks, adamhsparks@gmail.com with ERA5 support adapted from code by Emerson M. Del Ponte delponte@ufv.br from the r4pde.

Examples

# Example 1: Get weather for a date range
# IRRI Zeigler Experiment Station, wet season 2000
wth_range <- get_wth(
  lonlat = c(121.255669, 14.167425),
  dates = c("2000-06-30", "2000-12-31")
)
head(wth_range)
#> Key: <YYYYMMDD>
#>      YYYYMMDD   DOY  TEMP  TMIN  TMAX  RHUM  RAIN      LAT      LON
#>        <IDat> <int> <num> <num> <num> <num> <num>    <num>    <num>
#> 1: 2000-06-30   182 26.60 23.94 29.89 85.99 11.94 14.16742 121.2557
#> 2: 2000-07-01   183 25.50 24.26 27.39 90.80 22.45 14.16742 121.2557
#> 3: 2000-07-02   184 26.45 24.07 30.09 85.74 14.87 14.16742 121.2557
#> 4: 2000-07-03   185 25.88 24.82 27.53 93.76 19.77 14.16742 121.2557
#> 5: 2000-07-04   186 26.12 24.97 28.08 91.53 12.15 14.16742 121.2557
#> 6: 2000-07-05   187 26.28 25.51 27.70 91.45 21.90 14.16742 121.2557
nrow(wth_range) # Number of days
#> [1] 185

# Example 2: Get weather by start date and duration
# 120-day window from June 30, 2000
wth_duration <- get_wth(
  lonlat = c(121.255669, 14.167425),
  dates = "2000-06-30",
  duration = 120L
)
nrow(wth_duration) # Should be 120 rows
#> [1] 120

# Example 3: Verify both methods return equivalent data
# (within floating-point precision)
all.equal(wth_range[1:120], wth_duration)
#> [1] "Attributes: < Component \"POWER.Dates\": 1 string mismatch >"

# Example 4: Use retrieved weather in disease model
wth <- get_wth(
  lonlat = c(121.255669, 14.167425),
  dates = "2000-06-30",
  duration = 120L
)

# Run bacterial blight model
bb_sim <- bacterial_blight(wth, emergence = "2000-07-01")
#> Error in .window_weather(wth = wth, emergence = emergence, duration = duration): Insufficient weather data to run 120 days after emergence
head(bb_sim)
#> Error: object 'bb_sim' not found
plot(
  bb_sim$simday,
  bb_sim$intensity,
  type = "l",
  xlab = "Days After Emergence",
  ylab = "Disease Intensity"
)
#> Error: object 'bb_sim' not found

# Example 5: Multiple locations - compare disease risk
locations <- data.frame(
  name = c("IRRI_Laguna", "BRRI_Gazipur", "CIMMYT_Mexico"),
  lon = c(121.255669, 90.2167, -99.1828),
  lat = c(14.167425, 24.0028, 19.1296)
)

# Get weather for each location
wth_list <- lapply(
  1:nrow(locations),
  function(i) {
    get_wth(
      lonlat = c(locations$lon[i], locations$lat[i]),
      dates = "2000-06-30",
      duration = 120L
    )
  }
)
names(wth_list) <- locations$name

# Run leaf blast model at each location
lb_results <- lapply(
  wth_list,
  function(w) leaf_blast(w, emergence = "2000-07-01")
)
#> Error in .window_weather(wth = wth, emergence = emergence, duration = duration): Insufficient weather data to run 120 days after emergence

# Compare AUDPC (disease intensity summary)
audpc_comparison <- data.frame(
  location = names(lb_results),
  AUDPC = sapply(lb_results, function(x) attr(x, "AUDPC"))
)
#> Error: object 'lb_results' not found
audpc_comparison
#> Error: object 'audpc_comparison' not found

# Example 6: Cross-year comparison (same location, different years)
years <- 2000:2002
wth_by_year <- lapply(
  years,
  function(y) {
    get_wth(
      lonlat = c(121.255669, 14.167425),
      dates = paste0(y, "-06-30"),
      duration = 120L
    )
  }
)
names(wth_by_year) <- as.character(years)

# Run model for each year
bb_by_year <- lapply(
  wth_by_year,
  function(w) bacterial_blight(w, emergence = unique(w$YYYYMMDD)[1])
)

# Compare disease progression across years
plot_data <- data.table::rbindlist(lapply(
  seq_along(bb_by_year),
  function(i) {
    data.frame(
      year = names(bb_by_year)[i],
      simday = bb_by_year[[i]]$simday,
      intensity = bb_by_year[[i]]$intensity
    )
  }
))
plot(
  plot_data$simday,
  plot_data$intensity,
  col = as.factor(plot_data$year),
  pch = 16
)
legend(
  "topleft",
  legend = names(bb_by_year),
  col = seq_along(bb_by_year),
  pch = 16
)


# Example 7: Error handling
if (FALSE) { # \dontrun{
# This will fail (invalid coordinates)
get_wth(
lonlat = c(200, 95),  # Out of range
dates = "2000-06-30",
duration = 120
)
# Error: Longitude must be in [-180, 180], got 200

# This will fail (invalid date format)
get_wth(
lonlat = c(121.255669, 14.167425),
dates = "30-06-2000",  # Wrong format
duration = 120
)
# Error: dates[1] must be an ISO date ('YYYY-MM-DD')

# This will fail (missing date specification)
get_wth(
lonlat = c(121.255669, 14.167425),
dates = "2000-06-30"  # No end date or duration
)
# Error: Provide dates as c(start, end) or supply duration
} # }

# Example 8: ERA5 data
# Get ERA5 data in the same format as POWER
wth_era5 <- get_wth(
  lonlat = c(121.255669, 14.167425),
  dates = "2000-06-30",
  duration = 120L,
  source = "era5"
)
head(wth_era5)
#> Key: <YYYYMMDD>
#>      YYYYMMDD   DOY     TEMP  TMIN  TMAX     RHUM  RAIN   LAT    LON
#>        <IDat> <int>    <num> <num> <num>    <num> <num> <num>  <num>
#> 1: 2000-06-30   182 26.82917  24.9  30.0 89.37500  10.2 14.25 121.25
#> 2: 2000-07-01   183 27.02083  25.3  30.2 87.00000  42.4 14.25 121.25
#> 3: 2000-07-02   184 26.65000  24.1  29.5 89.62500  26.2 14.25 121.25
#> 4: 2000-07-03   185 26.23333  25.4  27.7 89.00000  25.7 14.25 121.25
#> 5: 2000-07-04   186 25.72083  25.0  26.4 92.37500  42.9 14.25 121.25
#> 6: 2000-07-05   187 26.34167  25.5  27.5 90.95833  31.9 14.25 121.25