Function Library

[ilonah22]

@julieshortridge suggested that it may be helpful for us to put some information about all of the functions that we have been using into one place. So, we are going to start adding some of those here, that way we can make sure functions aren't being duplicated and we can take suggestions on how to make them more efficient.

• bf_event_identifier - finds all baseflow events for a given time series (note: this does NOT retrieve data, it expects a properly formatted dataframe with flow timeseries).
• bf_event_stats - performs statistical analysis on an event to determine R, slope, and p-value, etc.
• trim_event_buffer
• trim_event_mk(event_df, alpha = 0.10, min_len_mk = 3, flag_len = 5): take a dataframe with a single event in it, iterates through trimming in order to arrive at an event with a Mann-Kendall p-value that is greater than the given threshold.
  • event_df
  • alpha
  • min_len_mk
  • flag_len

[ilonah22]

calc_event_stats

cal_event_stats() calculates the AGWRC for an event using the lm(log(Flow)~Date) method that we have been using.

source("https://raw.githubusercontent.com/HARPgroup/baseflow_storage/refs/heads/ih_function_cleanup/calc_event_stats.R")

calc_event_stats <- function (data, event_num, dAGWRmax, dAGWRmin){
require(sqldf)
  
source("https://raw.githubusercontent.com/HARPgroup/baseflow_storage/refs/heads/ih_function_cleanup/bf_event_stats.R")
  
sqldf_query <- paste0(
  "select * from data 
    where GroupID = ", event_num, " and
    AGWR < 1 and
    delta_AGWR < ", dAGWRmax," and
    delta_AGWR > ", dAGWRmin,"
    ")

event_data <- sqldf(sqldf_query)

values <- bf_event_stats(event_data)

new_df <- data.frame(event_data, calc_AGWRC = values$AGWRC, R_squared = values$R_squared)

return(new_df)

}

Notes

• The function should not be finding and analyzing data at the same time.
  • Call separate function to calculate AGWRC and R-Squared
• Change AGWR to AGWRC
• What should the output look like?
  • AGWRC and R-Squared columns added to input data frame?
  • Dataframe with only event_num, AGWRC, and R-Squared?
  • One AGWRC value and one R-Squared value for the input data?

[rburghol]

Thanks @ilonah22 for hopping on this. And thank @julieshortridge I think this is a super idea.

Ilonah:

• FWIW, see my comments regarding maybe (up to you) creating a sub-function called bf_event_stats that is called by calc_event_stats in order to not break other scripts that may expect calc_event_stats -- in BF Event post-processing #33 (comment)
• Also, when you are finished with the function doc, please migrate it from the comment above into the body of the issue so that we can know that it is finalized.

[ilonah22]

bf_event_stats

bf_event_stats - performs statistical analysis on an event to determine R, slope, and p-value, etc.

bf_event_stats currently only calculates AGWRC and R_squared for the data, since that was the main function of calc_event_stats.

source("https://raw.githubusercontent.com/HARPgroup/baseflow_storage/refs/heads/ih_function_cleanup/bf_event_stats.R")

bf_event_stats <- function(data, flow_col="Flow", date_col="Date"){

  data[[date_col]]<-as.Date(data[[date_col]])
  
  # Create lm of data
  logFlow_lm <-lm(log(data[[flow_col]]) ~ data[[date_col]])
  event_sum <- summary(logFlow_lm)
  
  # Assign AWGRC and R-squared
  AGWRC <- exp(event_sum$coefficients[[2,1]])
  R_squared <- event_sum$r.squared
  
  return(list(AGWRC = AGWRC, R_squared = R_squared))
  
}

Notes

• Thoughts on the format of column name inputs? I wasn't sure what the best way would be. I made them variable, but set default to what most of our data uses.
• Add other statistical analyses
  • p-value
  • mk?
• As it is right now, it should work with any dataset with a flow and date column (does not require GroupID). But, when you use it make sure the dataset isn't too large, I tried it with all CS flow data and it crashed R.
• Update 12/9 - I was getting R^2=1 for every event, so I had to add in a line to make sure date was in date format.

[rburghol]

@ilonah22 thie is great to see and I like your approach to arguments and their defaults. Simple and clean.