getArrhenius: Estimate Ea value and time varying Rb from temperature and...
In dpabon/ecofunr: Deriving Ecosystem Functional Properties

Description Usage Arguments Details Author(s)

Function to determine the temperature sensitivity for an Arrhenius-like model and time varying basal efflux (Rb(i)) from a given temperature and efflux (usually respiration) time series according the principle of 'SCAle dependent Parameter Estimation, SCAPE' (Mahecha et al. 2010).

getArrhenius(
  temperature,
  respiration,
  sf,
  fborder = 30,
  M = -1,
  nss = 0,
  method = "Fourier",
  weights = NULL,
  lag = NULL,
  gapFilling = TRUE,
  doPlot = FALSE
)

`temperature`	numeric vector: temperature time series.
`respiration`	numeric vector: respiration time series.
`sf`	numeric: sampling rate, number of measurements (per day).
`fborder`	numeric: boundary for dividing high- and low-frequency parts (in days).
`M`	numeric vector: size of SSA window (in days).
`nss`	numeric vector: number of surrogate samples.
`method`	String: method to be applied for signal decomposition (choose from 'Fourier','SSA','MA','wavMODWT','Spline')
`weights`	numeric vector: optional vector of weights to be used for linear regression, points can be set to 0 for bad data points.
`lag`	numeric vector: optional vector of time lags between respiration and temprature signal.
`gapFilling`	Logical: Choose whether Gap-Filling should be applied.
`doPlot`	Logical: Choose whether Surrogates should be plotted.

Function to determine the temperature sensitivity (Ea value) and time varying basal efflux (Rb) from a given temperature and efflux (usually respiration) time series. The following Arrhenius model could be used to describe the respiration:

Resp(i) = R_b * exp(- Ea / (R T(i))),

where i is the time index. It has been shown, however, that this model is misleading when Rb is varying over time which can be expected in many real world examples (e.g. Sampson et al. 2008).

If Rb varies slowly, i.e. with some low frequency then the 'scale dependent parameter estimation, SCAPE' allows us to identify this oscillatory pattern. As a consequence, the estimation of Ea can be substantially stabilized (Mahecha et al. 2010). The model becomes

Resp(i) = R_b(i) * exp(- Ea / (R T(i))),

where Rb(i) is the time varying 'basal respiration', i.e. the respiration expected at Tref. The convenience function getArrhenius allows to extract the Ea value minimizing the confounding factor of the time varying Rb. Four different spectral methods can be used and compared. A surrogate technique (function by curtsey of Dr. Henning Rust, written in the context of Venema et al. 2006) is applied to propagate the uncertainty due to the decomposition.

The user is strongly encouraged to use the function with caution, i.e. see critique by Graf et al. (2011).

Fabian Gans, Miguel D. Mahecha, MPI BGC Jena, Germany, fgans@bgc-jena.mpg.de mmahecha@bgc-jena.mpg.de

dpabon/ecofunr documentation built on July 15, 2020, 12:58 p.m.