FingerPro is a collaborative R project aiming to solve and share all modelling concerns around the sediment fingerprinting technique. Join us and discover the insights of your data!!
The package provides users with tools to i) characterise you database ii) assist in the selection of the optimal tracers moving forward from the traditional tracer selection methods (also included), iii) extract the multiple and consensual solutions in your database and iv) unmix sediment samples to estimate the apportionment of the sediment sources. The package used state-of-the-art equations and techniques for rigorous unmixing, avoiding previous thoughts about only relying on model capacity.
Table of Contents
For additional details, please see the recently published FingerPro paper and the newly developed tools such as the Consensus Ranking (CR) and the newest developed Consistency based tracer selection (CTS) that includes:
- Full description of functions and how to use them
- Full description of equations
If you're working with stable isotopes or want to combine them with elemental tracers, check the latest published method conservative balance (CB) and its applicability in a real case study.
👉 Frequently Asked Questions ➡️ Check them!
# From CRAN (version 1.1)
install.packages("fingerPro")
library(fingerPro)
# From your computer (version 1.3)
Download the fingerPro_1.3.zip file from GitHub on your computer
setwd("C:/your/file/directory")
install.packages('fingerPro_1.3.zip', repos = NULL)
# From GitHub (version 1.3)
devtools::install_github("eead-csic-eesa/fingerPro", ref = "master", force = T)To use your own data is as easy as to follow the format supplied in the example data included in the fingerPro package
print (catchment)When using raw data, the following structure needs to be followed
- The first column must be numeric corresponding to the sample ID
- The second column corresponds to the different sources
head (catchment[,c(1:6)])
id Land_Use Pbex K40 Bi214 Ra226
1 42665 AG 9.48 494 31.6 32.9
2 42666 AG 19.12 470 32.2 35.1
3 42667 AG 22.62 513 31.1 28.6
4 42694 AG 24.08 587 32.2 32.9
5 42741 AG 3.38 567 28.2 29.4
6 42770 AG 0.32 586 29.9 31.6- Your mixtures must be located in the last rows with the same name in the 2nd column and different ID (column 1)
tail (catchment[,c(1:6)])
id Land_Use Pbex K40 Bi214 Ra226
18 42707 PI1 33.5000 530.00 26.0000 24.0000
19 42807 SS 0.0000 478.00 26.1000 24.3000
20 42811 SS 0.0000 736.00 26.2000 28.2000
21 42812 SS 0.0000 607.00 26.1500 26.2500
22 428112 SS 0.0000 613.07 26.4115 26.5125
23 42745 mix.sample 24.8258 460.56 27.0680 27.1690If you only have mean and SD data, follow the format supplied in the Kamish dataset example
sources.file <- system.file("extdata", "Raigani.csv", package = "fingerPro")
data <- read.csv(sources.file)
print(data)Once you have your data in the appropriate format, load it to your global environment
setwd("C:/Users/.....")
data <- read.table("your dataset.csv", header = T, sep = ',')The following example displays all the basic commands available in the package to display informative graphs.
If you want to use your own data, see the previous section preparing your data
#Load the dataset called "catchment"
data <- catchmentboxPlot(data, tracers = 1:3, ncol = 3 ,colors = c("#993300", "#33CC00", "#336600", "#9933CC", "#0000FF"))
correlationPlot(data, columns = 1:6, mixtures = T, colors = c("#993300", "#33CC00", "#336600", "#9933CC", "#0000FF"))
LDAPlot(data[, c(1:10)], text = T, P3D = F, interactive = F, colors = c("#993300", "#33CC00", "#336600", "#9933CC"))
LDAPlot(data[, c(1:10)], text = T, P3D = T, interactive = F, colors = c("#993300", "#33CC00", "#336600", "#9933CC"))
LDAPlot(data[, c(1:10)], text = T, P3D = T, interactive = T, colors = c("#993300", "#33CC00", "#336600", "#9933CC"))
PCAPlot(data, components = c(1:2), colors = c("#993300", "#33CC00", "#336600", "#9933CC", "#0000FF"))
Moving forward from the traditionally implemented tracer selection methods proven wrong by recent research, this section explains how to implement state-of-the-art approaches to extract individual tracer information and multiple solutions to assist you in this crucial step.
Major benefits:
- Understanding your dataset
- Are there specific relationships among my tracers?
- Does my dataset have multiple solutions?
- What's leading my model to its results?
- Agreement between different models (e.g. FingerPro & MixSIAR)
sources.file <- system.file("extdata", "Raigani.csv", package = "fingerPro")
data <- read.csv(sources.file)
#Compute the CI index and the individual tracer solutions
results_CI <- CI_Method(data, points = 3000, Means = T) # Means = F (When using raw data)
# Plot the individual tracers solution from the eight first tracers
Ternary_diagram(results_CI, tracers = c(1:6), n_row = 1, n_col = 6)
mixture <- tail(data, n = 1)
var <- grep("^D", colnames(data))
mixture <- mixture[-c(var)]
row.names(mixture) <- NULL
source <- head(data,-1)
sgeo <- source[, -1]
mgeo <- mixture[, -1]
# When using raw data
# sgeo <- inputSource(data)
# mgeo <- inputSample(data)
crgeo <- cr_ns(source=sgeo, mixture = mgeo, maxiter = 1000, seed = 1234567)
head(crgeo)
#RESULTS
tracer score
1 P 96.8
2 Li 96.1
3 Ba 96.0
4 K 95.6
. . .
. . .
. . .
31 Cu 0.7
32 Zn 0.7
33 Pb 0.6
34 V 0.1
# compute singles/pairs/triplets/quartets (depending on your source numbers 2-5 sources)
pgeo <- pairs(sgeo, mgeo, iter = 1000, seed = 1234567)
head(pgeo)
id w1 w2 w3 Dw1 Dw2 Dw3 cons Dmax
1 P Sr 0.5959349 0.20660832 0.197456756 0.05212140 0.05042700 0.04688297 1.000 0.05212140
2 Ba Sr 0.6317626 0.09009448 0.278142958 0.05230141 0.03341239 0.04975232 0.995 0.05230141
3 Sr Th 0.6673848 -0.02575151 0.358366677 0.05470869 0.04601870 0.05972316 0.275 0.05972316
4 Sr Ti 0.4723578 0.60848947 -0.080847215 0.05934662 0.06579869 0.04640424 0.038 0.06579869
5 Sr Mg 0.5115884 0.48090866 0.007502897 0.04083850 0.07169679 0.07250162 0.540 0.07250162
6 K Sr 0.6340332 0.08271014 0.283256639 0.05456926 0.07371975 0.07234518 0.851 0.07371975Thanks to this method, we can see the presence of multiple solutions that would otherwise go undetected. Now is the time to use all the accumulated expertise from statistical methods, lab work, and fieldwork to decide.
# Explore those pairs and multiple solutions (e.g. P-Sr and Ba-Sr)
# Solution 1
sol_1 <- pgeo[pgeo$id=="P Sr",]
ctsgeo_1 <- cts_3s(source = sgeo, mixture = mgeo, sol = c(sol_1$w1, sol_1$w2, sol_1$w3))
ctsgeo_1 <- ctsgeo_1 %>% right_join(crgeo, by = c("tracer"))
ctsgeo_1 <- ctsgeo_1[ctsgeo_1$err < 0.025 & ctsgeo_1$score > 80,]
# Solution 2
sol_2 <- pgeo[pgeo$id=="Ba Sr",]
ctsgeo_2 <- cts_3s(source = sgeo, mixture = mgeo, sol = c(sol_2$w1, sol_2$w2, sol_2$w3))
ctsgeo_2 <- ctsgeo_2 %>% right_join(crgeo, by = c("tracer"))
ctsgeo_2 <- ctsgeo_2[ctsgeo_2$err < 0.025 & ctsgeo_2$score > 80,]
print(cbind(ctsgeo_1, ctsgeo_2))
tracer err score tracer err score
Li 2.176625e-02 96.1 Ba 1.110223e-16 96.0
P 2.220446e-16 96.8 K 4.117438e-03 95.6
Sr 3.330669e-16 93.6 Li 1.541905e-02 96.1
Ca 1.337685e-02 93.0 Sr 1.998401e-15 93.6
data_sol_1 <- select(data, "id", "sources", "Ba", "Li", "K", "Sr", "DBa", "DLi", "DK", "DSr", "n")
data_sol_2 <- select(data, "id", "sources", "Li", "P", "Sr", "Ca", "DLi", "DP", "DSr", "DCa", "n")
# Let's unmix the multiple solutions
result_FP_1 <- unmix(data_sol_1, samples = 200, iter = 200, Means = T)
result_FP_2 <- unmix(data_sol_2, samples = 200, iter = 200, Means = T)
P_FP_1 <- plotResults(result_FP_1, y_high = 1, colors = c("#CC0000", "#33CCFF", "#9933CC"))
P_FP_2 <- plotResults(result_FP_2, y_high = 1, colors = c("#CC0000", "#33CCFF", "#9933CC"))
grid.arrange(P_FP_1, P_FP_2, ncol=2)
Also available on bilibili
Also available on bilibili
You can cite this package and the new developed tools on your work as:
Lizaga, I., Latorre, B., Gaspar, L., Navas, A., 2020. FingerPro: an R package for tracking the provenance of sediment. Water Resources Management 272, 111020. https://doi.org/10.1007/s11269-020-02650-0.
Lizaga, I., Latorre, B., Gaspar, L., Navas, A., 2020a. Consensus ranking as a method to identify non-conservative and dissenting tracers in fingerprinting studies. Science of The Total Environment 720, 137537. https://doi.org/10.1016/j.scitotenv.2020.137537
Latorre, B., Lizaga, I., Gaspar, L., Navas, A., 2021. A novel method for analysing consistency and unravelling multiple solutions in sediment fingerprinting. Science of The Total Environment 789, 147804. https://doi.org/10.1016/j.scitotenv.2021.147804
Lizaga, I., Latorre, B., Gaspar, L., Navas, A., 2022. Combined use of geochemistry and compound-specific stable isotopes for sediment fingerprinting and tracing. Science of The Total Environment 832, 154834. https://doi.org/10.1016/j.scitotenv.2022.154834
and also refer to the code as:
Lizaga I., Latorre B., Gaspar L., Navas A., (2018) fingerPro: An R package for sediment source tracing, https://doi.org/10.5281/zenodo.1402029.
This software has been improved by the questions, suggestions, and bug reports of the user community. If you have any comments, please use the Issues page or report them to [email protected].
- Combining geochemistry and isotopic tracers
- New tools for understanding individual tracers and tracer selection methodologies
- Sediment source fingerprinting in Glacial Landscapes, Svalbard, Peruvian Andes and Antarctica
- Agricultural Cycle influence in sediment and pollutant transport
- Changes in source contribution during an exceptional storm event and before and after the event
- Sediment source fingerprinting in desert environments
- Sediment source fingerprinting to track pollutants in mountainous fluvial environment, mining areas and agroecosystems
- Testing FingerPro model with artificial samples
- Particle size effect