im_ratio_dB: "Calculates the intermittancy ratio (in percentage) and...

In MarloesEeftens/NOISEtools: Helpful tools for environmental noise measurements and research.

Description

Given a time series of sound levels as a numeric vector or variable within a dataframe, this function calculates the intermittancy ratio based on the measured sound levels.

Usage

im_ratio_dB(x,L,C,timevar,plot_filename)

Arguments

x	A numeric vector or R dataframe that includes the variable for which the intermittancy ratio should be calculated.
L	If x is a dataframe, L is the name of the variable within the dataframe which contains the sound level measurements for which IR should be calculated.
C	(Optional, defaults to C=3.) The "only free parameter in the definition of IR" (Wunderli, 2016). C is the difference between the average sound level Leq_T_tot and the threshold level at which we define an "event", or Leq_T_event. Setting C lower will result in a higher eventfullness.
timevar	(Optional, defaults to any variable of class "POSIXct", if present in dataframe specified in x.) This is the variable which contains the datetime values for the time series.
plot_filename	If a visual representation of the time series is required, you can specify here which folder and filename it can be saved to.

Details

Please note that a noise "event" (see below) can last more than 1 second. So the number of events is not equal to the number of values above the threshold, but equal to the number of times the time series goes above the K threshold line.

Value

The result is a named vector with 7 different statistics in it: 1) N_obs, the number of observations in the time series; 2) LEQ_T_tot, the average sound level during the measurement; 3) C, The value of the offset used to determin the exceedance threshold for events; 4) K, the threshold whose exceedance defines an "event"; 5) N_events, the number of events (see also details); 6) LEQ_T_events, the average amount of sound energy in dB[A] which can be attributed to events; 7) IM_ratio, the intermittancy ratio, a numeric value without unit, see reference below. There is an option to export a visual representation of the time series into a chosen plot_filename.

Note

The intermittancy ratio was conceived to be calculated over a time series of sound measurements with measurement interval 1 second. In the referenced paper, separate statistics were calculated over each hour of the day (e.g. for 1-second data from 12:00:00 to 12:59:59, then from 13:00:00 to 13:59:59, etc). While it is possible to calculate the intermittancy ratio measured over measurement time series with a different interval than 1 second, be conscious that this is not how the statistic was conceived. It is also possible to calculate the imtermittancy ratio over shorter or longer periods than 1 hour, but be aware that during long measurements (e.g. over the course of a 24-hour day), the "background" noise level (above which the threshold for events "K" is calculated) changes, which prevents the accurate identification of events. Calculating an IM ratio over shorter periods will work as well, but be aware of the random noise (literally!) in short time series: short time series give you unstable estimates of both Leq and the IM ratio. This is demonstrated in the 3rd example below: in this particular example, the estimation of IM ratio stabilizes after approximately 15 minutes of measuring at the same spot.

Author(s)

Marloes Eeftens, [e-mail]

References

Jean Marc Wunderli, Reto Pieren, Manuel Habermacher, Danielle Vienneau, Christian Cajochen, Nicole Probst-Hensch, Martin Röösli, Mark Brink. Intermittency ratio: A metric reflecting short-term temporal variations of transportation noise exposure Journal of Exposure Science and Environmental Epidemiology (2016) 26, 575–585 (2016) doi:10.1038/jes.2015.56

Examples

#Write the example .csv to a folder of your choice (please change pathname!):
my_filename<-"V:/EEH/R_functions/NOISEtools/data/sentry_example.csv"
write.table(sentry_example,file=my_filename,sep="\t",quote=FALSE,row.names=FALSE)
#Import using the import function:
my_sentry_file<-import_sentry(filename=my_filename)

#Calculate the intermittancy ratio over the variable "LEQ"
im_ratio_dB(my_sentry_file$LEQ)

#Calculate and plot the intermittancy ratio for 6 different levels of C (requires library ggplot2!).
for (i in seq(1:6)){
  stats<-im_ratio_dB(x=my_sentry_file,L="LEQ",C=i,timevar="PosixTime",plot_filename=paste0("V:/EEH/R_functions/NOISEtools/plots/ir_",i,".png"))
  print(paste0("C=",i,", Intermittancy Ratio=",round(stats[["IM_ratio"]])))
}

#See how the intermittancy ratio develops as we measure longer at the same spot.
irs_cum=list()
for(i in seq(1:1800)){irs_cum[[i]]<-im_ratio_dB(x=my_sentry_file[1:i,],timevar="PosixTime",L="LEQ")}
irs_cum_df<-as.data.frame(do.call(rbind,irs_cum))
ggplot(data=irs_cum_df,aes(x=N_obs,y=IM_ratio))+
  geom_point()+
  geom_hline(yintercept=tail(irs_cum_df$IM_ratio,1),color="blue")

MarloesEeftens/NOISEtools documentation built on Dec. 14, 2019, 5:54 p.m.