Heatwave calculation based on maximum temperature in R

Question

A heatwave is defined if the maximum temperature at a meteorological station is 3 °C or more than the normal temperature consecutively for 3 days or more. I have calculated the daily average (daily normal) from multiple-year daily maximum temperature data like

df <- data.frame("date"= seq(from = as.Date("1970-1-1"), to = as.Date("2000-12-31"), by = "day"),
                        "MaxT" = runif(length(seq.Date(as.Date("1970-1-1"), as.Date("2000-12-31"), "days")), 20, 40))

df$day <- format(df$date, format='%m-%d')

daily_mean <- aggregate(MaxT ~ day, data=df, FUN=mean)

Now it has to be matched with every year's daily maximum temperature and identify the dates when the maximum temperature is 3 °C or more than the normal daily temperature consecutively for 3 days or more. Those events will be considered as heatwaves. My question is how to implement it in R?

The yearly data that you will receive would be in same format as `daily_mean` or it would be of the format as in `df` ? — Ronak Shah, Apr 13 '20 at 07:19
I want the yearly number of such events and the dates (or starting dates) of those events. — UseR10085, Apr 13 '20 at 07:22

Ronak Shah · Accepted Answer · 2020-04-13T08:42:06.020

Here is a way using dplyr and zoo's rollapplyr.

library(dplyr)

df_out <- df %>%
           left_join(daily_mean %>% rename(mean_temp = MaxT), by = 'day') %>% 
            mutate(is_heatwave = zoo::rollapplyr(MaxT > (mean_temp + 3), 
              3, all,fill = NA))

some instances of heatwave :

df_out[31:50, ]

#         date     MaxT   day mean_temp is_heatwave
#31 1970-01-31 26.31675 01-31  28.31451       FALSE
#32 1970-02-01 22.05946 02-01  29.83059       FALSE
#33 1970-02-02 34.22469 02-02  29.84562       FALSE
#34 1970-02-03 33.03264 02-03  29.87919       FALSE
#35 1970-02-04 36.62357 02-04  31.50603        TRUE
#36 1970-02-05 29.82134 02-05  30.22581       FALSE
#37 1970-02-06 28.13625 02-06  29.64073       FALSE
#38 1970-02-07 29.95754 02-07  29.54277       FALSE
#39 1970-02-08 21.40026 02-08  30.96619       FALSE
#40 1970-02-09 33.10983 02-09  28.16146       FALSE
#41 1970-02-10 30.87346 02-10  29.37693       FALSE
#42 1970-02-11 31.08721 02-11  28.89930       FALSE
#43 1970-02-12 27.34925 02-12  29.27882       FALSE
#44 1970-02-13 31.88582 02-13  29.35825       FALSE
#45 1970-02-14 30.05155 02-14  28.24995       FALSE
#46 1970-02-15 35.07049 02-15  29.02716       FALSE
#47 1970-02-16 39.49029 02-16  32.75644       FALSE
#48 1970-02-17 37.41917 02-17  31.44022        TRUE
#49 1970-02-18 36.03564 02-18  29.56212        TRUE
#50 1970-02-19 36.48052 02-19  30.18766        TRUE

TRUE values are where heatwave was present. As we can see in row 33, 34 and 35 we had 3 consecutive days when MaxT was greater than mean_temp by more than 3 degrees. Similarly, we can verify for other days.

To get yearly heatwave occurrences, we can do :

df_year <- df_out %>%
             group_by(year = format(date, "%Y")) %>%
             summarise(total_heat = with(rle(is_heatwave), 
                                         sum(values, na.rm = TRUE)))

and sum(df_year$total_heat) would give overall count.

This answers one of my question to have the dates of such event. Can we also have the count of such event per year and overall heatwave count? — UseR10085, Apr 13 '20 at 08:17
If the heatwave continues for consecutively for 3 days or more should be counted as one event. I think your calculation leads to overestimation of heatwaves. Can you please check that? — UseR10085, Apr 13 '20 at 08:31
will you please visit [this](https://stackoverflow.com/questions/69535937/how-to-calculate-a-condition-using-dplyr)? — UseR10085, Oct 14 '21 at 09:12

Heatwave calculation based on maximum temperature in R

1 Answers1

Linked