heatwaveR issues

Author

AJ Smit

Published

March 29, 2023

# devtools::install_github("robwschlegel/heatwaveR")
library(tidyverse)
library(heatwaveR)
# session_info()

# heatwaves
clm1 <- ts2clm(sst_WA, climatologyPeriod = c("1982-01-01", "2011-12-31"))

Applying detect_event(..., protoEvent = FALSE) (the default) returns protoEvents as part of the climatology even if they were not requested. The protoEvents must only be returned when protoEvents = TRUE.

# heatwaves
ev1 <- detect_event(clm1)

When detect_event(..., protoEvent = TRUE), a climatology with the protoEvents is returned without the associated detected events (expected behaviour upheld).

# heatwaves
ev2 <- detect_event(clm1, protoEvents = TRUE)

Below, swithing on climatology = TRUE returns the events in a long dataframe, but it is not actually a climatology. From the helpfile:

“If set to TRUE, this function will return a list of two dataframes, same as detect_event. The first dataframe climatology, contains the same information as found in detect_event, but with the addition of the daily intensity (anomaly above seasonal doy threshold) and category values.”

Below, the ‘climatology’ returned by category() is a truncated climatology since dates and associated temperatures on the dates when events were not detected are not present. The climatology returned by detect_event() has ALL the data, from the first day of the raw data time series right through to the last. Either the help file must be updated, or the full climatology must be returned.

cat1 <- category(ev1, name = "WA", climatology = TRUE)

There is an issue with detect_event(..., categories = TRUE) when fed a climatology created from a temperature time series which does not have the standard name, temp:

sst <- sst_Med |> 
  rename(temperature = temp)

clm2 <- ts2clm(sst, y = temperature, climatologyPeriod = c("1982-01-01", "2011-12-31"))

# events are named and seasons are present, but columns `p_moderate`, `p_strong`, `p_severe` and `p_extreme` are empty
ev3 <- detect_event(clm2, y = temperature, categories = TRUE)

# here the expected behaviour the above columns is observed
ev4 <- detect_event(clm2, y = temperature)
cat2 <- category(ev4, y = temperature)

# using the climatology that has the name `temp` works fine when the categories are requested as part of the `detect_event()` function
ev5 <- detect_event(clm1, categories = TRUE)

Lastly, for the S = TRUE switch in category(), what happens when the data straddle the equator? Also, there is no way to specify the S or N hemispheres when categories are requested as part of detect_event() so this should probably be added.

Reuse

CC BY-NC-SA 4.0

Citation

BibTeX citation:

@online{smit,_a._j.,
  author = {Smit, A. J., and Smit, AJ},
  title = {heatwaveR Issues},
  date = {},
  url = {http://tangledbank.netlify.app/vignettes/heatwaveR_issues.html},
  langid = {en}
}

For attribution, please cite this work as:

Smit, A. J., Smit A heatwaveR issues. http://tangledbank.netlify.app/vignettes/heatwaveR_issues.html.

--- author: "AJ Smit" date: "`r Sys.Date()`" title: "heatwaveR issues" bibliography: ../references.bib csl: ../marine-biology.csl format: html: default-image-extension: svg code-fold: false toc-title: "On this page" standalone: true toc-location: right page-layout: full smooth-scroll: true anchor-sections: true number-sections: true --- ```{r} #| message: false #| warning: false # devtools::install_github("robwschlegel/heatwaveR") library(tidyverse) library(heatwaveR) # session_info() ``` ```{r} # heatwaves clm1 <- ts2clm(sst_WA, climatologyPeriod = c("1982-01-01", "2011-12-31")) ``` Applying `detect_event(..., protoEvent = FALSE)` (the default) returns protoEvents as part of the climatology even if they were not requested. The protoEvents must only be returned when `protoEvents = TRUE`. ```{r} # heatwaves ev1 <- detect_event(clm1) ``` When `detect_event(..., protoEvent = TRUE)`, a climatology with the protoEvents is returned without the associated detected events (expected behaviour upheld). ```{r} # heatwaves ev2 <- detect_event(clm1, protoEvents = TRUE) ``` Below, swithing on `climatology = TRUE` returns the events in a long dataframe, but it is not actually a climatology. From the helpfile: > "If set to TRUE, this function will return a list of two dataframes, same as `detect_event`. The first dataframe climatology, contains the same information as found in `detect_event`, but with the addition of the daily intensity (anomaly above seasonal doy threshold) and category values." Below, the 'climatology' returned by `category()` is a truncated climatology since dates and associated temperatures on the dates when events were not detected are not present. The climatology returned by `detect_event()` has ALL the data, from the first day of the raw data time series right through to the last. Either the help file must be updated, or the full climatology must be returned. ```{r} cat1 <- category(ev1, name = "WA", climatology = TRUE) ``` There is an issue with `detect_event(..., categories = TRUE)` when fed a climatology created from a temperature time series which does not have the standard name, `temp`: ```{r} sst <- sst_Med |> rename(temperature = temp) clm2 <- ts2clm(sst, y = temperature, climatologyPeriod = c("1982-01-01", "2011-12-31")) # events are named and seasons are present, but columns `p_moderate`, `p_strong`, `p_severe` and `p_extreme` are empty ev3 <- detect_event(clm2, y = temperature, categories = TRUE) # here the expected behaviour the above columns is observed ev4 <- detect_event(clm2, y = temperature) cat2 <- category(ev4, y = temperature) # using the climatology that has the name `temp` works fine when the categories are requested as part of the `detect_event()` function ev5 <- detect_event(clm1, categories = TRUE) ``` Lastly, for the `S = TRUE` switch in `category()`, what happens when the data straddle the equator? Also, there is no way to specify the S or N hemispheres when categories are requested as part of `detect_event()` so this should probably be added.