Summarise sub-daily weather into index values, in the correct order

Computes the Mosquito Weather Index at the resolution of the supplied observations and only then summarises it within each group. This is the order the index requires, and getting it the wrong way round is the most common way to misuse it.

Usage

mwi_summarise(
  temperature,
  humidity,
  wind_speed,
  by = NULL,
  statistic = c("max", "mean", "min"),
  units = c("km/h", "m/s"),
  na.rm = FALSE
)

Arguments

temperature

Numeric vector of air temperatures in degrees Celsius.

humidity

Numeric vector of relative humidities in percent (0-100).

wind_speed

Numeric vector of wind speeds, in the units given by units.

by

Grouping vector the same length as the weather inputs, typically a Date or a day-of-year. Values are summarised within each unique element. Use NULL to summarise everything into a single value.

statistic

Summary to apply within each group: "max" (the default), "mean", or "min".

units

Units of wind_speed: "km/h" (the default) or "m/s".

na.rm

Whether to drop missing index values within each group. If FALSE (the default), a group containing any NA returns NA.

Value

A data frame with one row per group and columns by, mwi and n (the number of observations contributing to the group). If by is NULL, a one-row data frame with by = NA.

Details

The index is a product of non-linear response functions, so it does not commute with averaging: mwi(mean(temperature), mean(humidity), mean(wind)) is not mean(mwi(temperature, humidity, wind)). The discrepancy is severe with daily maxima, because the daily maximum of relative humidity and the daily maximum of temperature usually fall many hours apart, so an index computed from them describes conditions that never co-occurred.

mwi_summarise() therefore takes sub-daily observations, evaluates mwi() on each one, and reduces the results with statistic within each level of by. Empirically the maximum is the more informative summary: it captures whether the period contained a window of favourable conditions, rather than how favourable the period was on average.

See also

mwi() for the index itself.

Examples

# One synthetic day of hourly weather
hours <- 0:23
temp <- 18 + 8 * sin((hours - 6) / 24 * 2 * pi)
rh <- 75 - 15 * sin((hours - 6) / 24 * 2 * pi)
wind <- rep(5, 24)

mwi_summarise(temp, rh, wind, by = rep("2019-07-15", 24))
#>           by       mwi  n
#> 1 2019-07-15 0.5657766 24

# The wrong order gives a different answer
mwi(max(temp), max(rh), max(wind))
#> [1] 0.7272727