In lsigut/openeddy: Post-process eddy covariance data with ease

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "README-"
)

openeddy

Overview

The openeddy package is a software infrastructure for eddy covariance data post-processing. It provides tools for handling data with attached metadata, setting up quality control scheme and summarizing and plotting the results. It is aligned with REddyProc package that provides methods for uStar-filtering, gap-filling, and flux-partitioning. Thus the combined use of openeddy and REddyProc allows to run the whole eddy covariance post-processing chain.

All openeddy functions are extensively documented. The help file of the given function with the arguments description, details section, and examples can be viewed using ?function_name. To see the most common openeddy use-cases, you can explore the commented interactive openeddy tutorials. The applications are also summarized in the form of a poster: https://zenodo.org/record/8159040. For the example of the whole eddy covariance post-processing chain see EC_workflow.

Installation

1. Install devtools package if not available yet.

   r install.packages("devtools")

2. Install openeddy

   r devtools::install_github("lsigut/openeddy")

Extended Example

An extended example of applied openeddy software infrastructure is available at:

https://github.com/lsigut/EC_workflow

Short Example

Demonstration of selected generally applicable openeddy functions.

library(openeddy)
library(REddyProc)
library(ggplot2)

Example data from REddyProc package do not include statistics extracted from raw data or quality control (QC) flags. Notice that units are included.

data(Example_DETha98)
str(Example_DETha98[1:4])
DETha98 <- fConvertTimeToPosix(Example_DETha98, 'YDH', 
                               Year = 'Year', Day = 'DoY', Hour = 'Hour')

Certain variable names are expected by openeddy, thus DETha98 variable names need to be renamed. Net ecosystem exchange (NEE) is already filtered based on QC information (qc_NEE) that is not included. Though flag 2 values can be recreated, respective NEE values are missing and flag 1 cannot be resolved.

rename <- names(DETha98) %in% c("DateTime", "Rg")
names(DETha98)[rename] <- c("timestamp", "GR")
DETha98$qc_NEE <- ifelse(is.na(DETha98$NEE), NA, 0)

Application of three general filters is presented.

1. Low covariance between vertical wind component and CO~2~ concentration ((CO~2~)) can be caused by frozen ultrasonic anemometer or problems with (CO~2~) measurements. Such cases are flagged and saved to qc_NEE_lowcov.
2. Runs with repeating values are a sign of malfunctioning equipment (qc_NEE_runs).
3. Spikes in low frequency data cause problems during gap-filling and should be excluded. Since DETha98 was already quality checked, the amount of detected spikes is limited. In order to correctly evaluate spikes, preliminary QC (qc_NEE_prelim) that combines available QC tests or filters should be produced and used in despikeLF.

DETha98$qc_NEE_lowcov <- 
  apply_thr(DETha98$NEE, c(-0.01, 0.01), "qc_NEE_lowcov", "between")
summary_QC(DETha98, "qc_NEE_lowcov")
DETha98$qc_NEE_runs <- flag_runs(DETha98$NEE, "qc_NEE_runs")
summary_QC(DETha98, "qc_NEE_runs")
DETha98$qc_NEE_prelim <- 
  combn_QC(DETha98, 
           c("qc_NEE", "qc_NEE_lowcov", "qc_NEE_runs"), 
           "qc_NEE_prelim")
DETha98$qc_NEE_spikesLF <- 
  despikeLF(DETha98, "NEE", "qc_NEE_prelim", "qc_NEE_spikesLF", 
            light = NULL)
summary_QC(DETha98, "qc_NEE_spikesLF")

The QC results can be summarized in tabular or graphical form using summary_QC. It is possible to summarize each filter independently or summarize the cumulative effect of applied filters. Note that the fraction of flagged records in this example is negligible as DETha98 data set was already quality checked.

summary_QC(DETha98, 
           c("qc_NEE", "qc_NEE_lowcov", "qc_NEE_runs", "qc_NEE_spikesLF"))
summary_QC(DETha98, 
           c("qc_NEE", "qc_NEE_lowcov", "qc_NEE_runs", "qc_NEE_spikesLF"),
           cumul = TRUE, plot = TRUE, flux = "NEE")

Although individual QC columns should be stored as they are useful to distinguish the reason why certain records were excluded, only the combined QC column (qc_NEE_composite) is usually used in further data processing and analysis.

DETha98$qc_NEE_composite <- 
  combn_QC(DETha98, 
           c("qc_NEE", "qc_NEE_lowcov", "qc_NEE_runs", "qc_NEE_spikesLF"), 
           "qc_NEE_composite")

Function plot_eddy is useful for visualization of the whole dataset including flux values, its respective QC flags and the most important micrometeorological parameters in monthly and weekly time resolution. Only a two week subset is presented here to limit the extent of output.

DETha98[, c("P", "PAR", "Rn")] <- NA
(varnames <- varnames(DETha98))
(units <- openeddy::units(DETha98))
sub <- DETha98$DoY >= 29 & DETha98$DoY < 43
DETha98_sub <- DETha98[sub, ]
openeddy::units(DETha98) <- units
plot_eddy(DETha98_sub, "NEE", "qc_NEE", "qc_NEE_composite", skip = "monthly",
          light = "GR")

In addition to actual despiking, despikeLF can be used also for visualization of the internally computed double-differenced time series in order to inspect selected 13 days blocks. See section Plotting in despikeLF help file for further description.

despikeLF_plots <- 
  despikeLF(DETha98, "NEE", "qc_NEE_prelim", "qc_NEE_spikesLF", 
            light = NULL, plot = TRUE)$plots
despikeLF_plots$`iter 1`$all$`1998-01-27 - 1998-02-08`

Contact

Ladislav Šigut: sigut.l\@czechglobe.cz

Global Change Research Institute CAS, Bělidla 4a, 603 00 Brno, CZ

References

Publication describing openeddy is not yet available. When describing the proposed quality control scheme, please refer to it as similar to:

McGloin, R., Sigut, L., Havrankova, K., Dusek, J., Pavelka, M., Sedlak, P., 2018. Energy balance closure at a variety of ecosystems in Central Europe with contrasting topographies. Agric. For. Meteorol. 248, 418-431. https://doi.org/10.1016/j.agrformet.2017.10.003

lsigut/openeddy documentation built on Aug. 5, 2023, 12:25 a.m.