Sunset-calc is a R shiny app to calculate the amount of carbon for TC, Swiss_3S, and Swiss4S protocols from raw files recorded using a commercial thermo-optical OC/EC analyser (Model 5L, Sunset Laboratory Inc., OR, United States). You can also calculate the EC recovery after WINSOC removal (EC yield) with the Swiss_3S protocol. The 'OC/EC yield' gives you all the amounts, yield, and charring for OC and EC. You can upload one or multiple files, however, each file must contain only one Sunset run. If you have multiple runs in a single txt file, please use the 'file splitter' first.
Program to calculate TC for TC files. You get a .zip file containing a .csv file with the amount of carbon (µg C) for each step and the total carbon as well as the name of the sample corresponding to the input file name. If you compare the OC calc result to a result calculated by another software, be aware of the filter area that you used.
Program to calculate S1, S2, and S3 OC and total for Sunset OC-removal (Swiss_3S) files. See Zhang et al. (2012) for details. If you use a modified protocol, you need to adjust the set time in seconds for each modified step. Please do not adjust the times if you don't know what you are doing. When you reload the app, the times will go back to the default value, which corresponds to the standard Swiss_3S. You get a .zip file containing a .csv file with the amount of carbon (µg C) for each step and the total carbon as well as the name of the sample corresponding to the input file name. If you compare the OC calc result to a result calculated by another software, be aware of the filter area that you used.
Program to calculate S1, S2, S3, S4, and total for Swiss_4S files. The calculation is basically the same as the OC calc for Swiss_3S, however, it is adjusted for the additional (EC) step at the end.
Program to calculate the amount of carbon in S1, S2, S3, and total amout of carbon in OC (Swiss_3S file). The amount of EC is calculated with the uploaded EC file. Additionally, the EC yield and charring is calculated using the Swiss_3S file. The result from each calculation is then used to perform the amount correction.
This app calculates the EC yield and charring for one or multiple OC removal runs with the Swiss_3S protocol. This app is useful for the optimisation (time, temperature) of the OC/EC separtion. Frequently, you will need perform the OC removal on multiple filters cuts to get enough EC for a radiocarbon measurement. Yield calc calculates EC yield and charring, removes potential outliers and gives you the result you need. A generic file is used for laser signal correction. Optionally, you can upload your own file (TC, EC, or Swiss_4S) for temperature dependent laser signal correction.
You will get a zip folder containing five files:
| File name | File extension | Description |
|---|---|---|
| mean-result | csv | Mean results when multiple files were uploaded |
| raw-result | csv | EC yield and charring from all uploaded filters |
| clean-result | csv | EC yield and charring results without outliers |
| stat-result | csv | Statistics |
| yield-calc-summary | Summary pdf with boxplots for EC yield and charring and summary table |
The file splitter splits a Sunset txt raw file with multiple runs in one to multiple files with one run. The app is no-frills; upload the file and get a zip file with each run in a single txt file. Result txt file nomenclature: [sample number]-[file name]-[sample name]-split.txt
Additional to splitting files, the file splitter also shows you a table with some relevant information regarding each file in a table, which you can copy and download.
The 4S/3S converter converts Sunset txt raw files from Swiss_4S runs to files with three steps only. This means, data recorded during the last step of the Swiss_4S protocol are deleted. Therefore, the output files of the 4S/3S converter are equivalent to Sunset txt raw files recorded during a Swiss_3S run and can be treated analogously with the programs 'OC calc', 'OC/EC yield' and 'Yield calc'. Output txt file nomenclature: [file name]-converted.txt
The 4S/3S converter accepts splitted files only. Multiple runs can be uploaded and converted simultaneously. The output is provided in a zip file containing a txt file for each run.
The Sunset calc app is made with shinydashboard, which contains the apps TC calc, OC calc, Swiss 4S calc, OC/EC yield, and Yield calc linked in the sidebar. Additionally, there is a file splitter app in the sidebar. The plots immediately shown after file upload are generated independently from the calculation in the app, the calculation takes place in a linked R script and is triggered by pressing the 'Calculate & Download' button. A busy indicator made with shinybusy appears when the server is busy.
First, the coefficients from NDIR calibration are calculated. The amount reflects the known amount of analyte (sucrose solution) added, the area is the calculated area with the code below. Please be aware that the values storred file are valid for the Sunset device at LARA and might be significantly different on an other device. The csv file is imported is assigned the variable NDIR_calib. A linear regression model using the lm() function is made and the coefficients stored to the variable coef. Using the coefficients, the currentCalConstant is calculated. This CalConstant is later used to for correction should the sample have used a different calibration constant (e.g. when measuring online with GIS/MICADAS). Initially, the calculated intercept was used, however, trials showed that the results match the Sunset Calc426 (program provided by Sunset for analysis) better when the intercept was set to zero.
# calculation of coefficients from NDIR calibration, intercept set to zero
NDIR_calib <- read.csv("NDIR-integrals-20200224-offline.csv", header = T)
intercept <- 0
calib <- lm(I(area - intercept) ~ 0+ amount, data = NDIR_calib)
coef <- as.data.frame(calib$coefficients)
coef <- rbind(intercept, as.data.frame(calib$coefficients))
currentCalConstant <<- (mean(NDIR_calib$CH4.area)-coef[1,])/coef[2,]
The uploaded Sunset raw file is imported to a data frame, a new column for time added and all unnecessary columns except time_s and CO2_ppm removed.
df <- as.data.frame(read.csv(file = filename, sep = ",", skip = 28, header = T ))
df$time_s <- seq(1:length(df$CO2_ppm))
df <-df[,c(21,16)]
A baseline correction is then performed:
df$CO2_ppm <- df$CO2_ppm-median(sort(df$CO2_ppm,decreasing=F)[1:length(which(df$CO2_ppm < 0))])
df$CO2_ppm[df$CO2_ppm < 0] <- 0
The calibration constant (CalConstant) value is imported from the file and the calibration constant factor (CalConstFactor) calculated. The CalConstant in a file could be different due to a new calibration, an old file, or an online measurement. Online measurements are different due to a different back pressure to the NDIR. Pressure as well as temperature affect CO2 measurements with NDIR (Yasuda et al., 2012).
CalConstant <- as.data.frame(read.csv(file = filename, sep = ",", skip = 18, header = F ))
CalConstant <- as.numeric(CalConstant[1,1])
CalConstFactor <- CalConstant/currentCalConstant
The local regression model is made, the CH4 area integrated and with that the calibration_peak_correction_factor calculated. Finally, the TC area is integrated and corrected with the calibration peak correction factor. Then, the amount in µg C is calculated using the calibration coefficients and corrected with the CalConstFactor.
colnames(df) = c("x", "y")
model<-loess(y~x, span=0.05, data=df)
mod.fun<-function(x) predict(model,newdata=x)
The script is shown for TC, but works in a similar fashion for OC and Swiss 4S.
#CH4 correction
CH4_area <- integrate(mod.fun,280,380)
#calibration peak correction factor with CH4
calibration_peak_correction_factor <- mean(NDIR_calib$CH4.area)/CH4_area$value
#Calculate area for each peak and total
#total carbon
total_area <- integrate(mod.fun,50,250)
total_area <- total_area$value*calibration_peak_correction_factor
amount.tc <- (total_area-coef[1,])/coef[2,]
amount.tc <<- amount.tc*CalConstFactor
The calculation code from above was wrapped into a function:
data.load.func = function(filename) {
#code from above here
}
This function is executed for each uploaded file:
filename <- input$fileUploaded$datapath
#file name for output
filename.text <<- input$fileUploaded$name
df.amount <- NULL
for (i in filename){
data.load.func(i)
df.amount <- rbind(df.amount, data.frame(amount.tc))
}
# combine file name with ouput data
df.amount <- cbind(filename.text,df.amount)
colnames(df.amount) <- c("sample name","TC (ug C)")
df.amount
The resulting df.amount is handled back to the shiny app for output.
The file splitter is simple: The file with multiple Sunset runs is uploaded, the script looks for the keyword Sample, with which every Sunset raw data file starts and splits into individual .txt files. Therefore, even when Sunset runs with different lengths are performed (e.g., TC followed by Swiss_4S), the files are split correctly into individual .txt files.
The server side of the oc_ec_yield_app.R contains the following script:
#run yields calc, oc-calc, and tc calc, then return the desired data
source("yields_calc_shiny.R", local = TRUE)
source("oc_calc_shiny.R", local = TRUE)
source("tc_calc_shiny.R", local = TRUE)
return(list(df.yield=df_raw, df.amount.oc=df.amount.oc, df.amount.tc=df.amount.tc))
As soon as the user presses the 'Calculate & Download' button, the uploaded OC data is run with the yields_calc_shiny.R and oc_calc_shiny.R script and the EC data is run in the tc_calc_shiny.R script. the return call gives the results from each calculation. The results are stored in the new data frame df.result. Using this data frame, the corrected OC and EC amounts are calculated. TCcalculated is the sum of OC and EC, ECcorr is the EC yield corrected amount of EC, and OCcorr is the EC yield corrected amount of OC. Finally, the data frame is sorted and column names applied.
The exported csv file contains the sample name for both the OC file as well as the EC file to avoid mistakes before EC yield and charring results as well as the raw and calculated amount of carbon for each fraction.
This app was created by Martin Rauber for LARA, the Laboratory for the Analysis of Radiocarbon with AMS at the University of Bern. The EC yield calculation script is part of COMPYCALC and adapted from there. Sunset calc is available online: 14c.unibe.ch/sunsetcalc. Please get in touch for any bug fixes and suggestions!
Sunset-calc is released under the MIT License.