partGLControl.RdDefault list of parameters for Lasslop 2010 daytime flux partitioning
For highest compatibility to the pvWave code of G.Lasslop
(used by first BGC-online tool)
see function partGLControlLasslopCompatible.
partGLControl(LRCFitConvergenceTolerance = 0.001,
nLRCFitConvergenceTolerance = 0.001,
nBootUncertainty = 30L, minNRecInDayWindow = 10L,
isAssociateParmsToMeanOfValids = TRUE,
isLasslopPriorsApplied = TRUE, isUsingLasslopQualityConstraints = FALSE,
isSdPredComputed = TRUE, isFilterMeteoQualityFlag = FALSE,
isBoundLowerNEEUncertainty = TRUE, fixedTRefAtNightTime = NA,
isExtendTRefWindow = TRUE, smoothTempSensEstimateAcrossTime = TRUE,
isNeglectPotRadForNight = FALSE, NRHRfunction = FALSE,
isNeglectVPDEffect = FALSE, isRefitMissingVPDWithNeglectVPDEffect = TRUE,
fixedTempSens = data.frame(E0 = NA_real_,
sdE0 = NA_real_, RRef = NA_real_),
replaceMissingSdNEEParms = c(perc = 0.2,
minSd = 0.7), neglectNEEUncertaintyOnMissing = FALSE,
minPropSaturation = NA, useNightimeBasalRespiration = FALSE)convergence criterion for rectangular light response curve fit. If relative improvement of reducing residual sum of squares between predictions and observations is less than this criterion, assume convergence. Decrease to get more precise parameter estimates, Increase for speedup.
convergence criterion for nonrectangular light response curve fit. Here its a factor of machine tolerance.
number of bootstrap samples for estimating uncertainty. Set to zero to derive uncertainty from curvature of a single fit
Minimum number of data points for regression
set to FALSE to associate parameters to the first record of the window for interpolation instead of mean across valid records inside a window
set to TRUE to apply strong fixed priors on LRC fitting. Returned parameter estimates claimed valid for some case where not enough data was available
set to TRUE to avoid quality constraints additional to Lasslop 2010
set to FALSE to avoid computing standard errors of Reco and GPP for small performance increase
set to TRUE to use only records where quality flag of meteo drivers (radiation, temperature, VPD) is zero, i.e. non-gapfilled for parameter estimation. For prediction, the gap-filled value is used always, to produce predictions also for gaps.
set to FALSE to avoid adjustment of very low uncertainties before day-Time fitting that avoids the high leverage those records with unreasonable low uncertainty.
if a finite value (degree Centigrade) is given, it is used instead of median data temperature as reference temperature in estimation of temperature sensitivity from night data
set to FALSE to avoid successively extending the night-time window in order to estimate a temperature sensitivity where previous estimates failed
set to FALSE to use independent estimates of temperature sensitivity on each windows instead of a vector of E0 that is smoothed over time
set to TRUE to not use potential radiation in determining night-time data.
deprecated: Flag if TRUE use the NRHRF
for partitioning; Now use lrcFitter = NonrectangularLRCFitter()
set to TRUE to avoid using VPD in the computations. This may help when VPD is rarely measured.
set to FALSE to avoid
repeating estimation
with isNeglectVPDEffect = TRUE trying to predict when VPD
is missing
data.frame
of one row or nRow = nWindow
corresponding to return value of partGLFitNightTimeTRespSens
While column RRef is used only as a prior and initial value for
the daytime-fitting and can be NA,
E0 is used as given temperature sensitivity and varied according
to sdE0 in the bootstrap.
parameters for
replacing missing standard deviation of NEE.
see replaceMissingSdByPercentage.
Default sets missing uncertainty to 20% of NEE but at least 0.7
flux-units (usually mumol CO2 / m2 / s).
Specify c(NA, NA) to avoid replacing missings in standard deviation of
NEE and to omit those records from LRC fit.
If set to TRUE: if there are records with missing uncertainty of NEE inside one window, set all uncertainties to 1. This overrules option replaceMissingSdNEEParms.
quality criterion for sufficient data in window. If GPP prediction of highest PAR of window is less than minPropSaturation * (GPP at light-saturation, i.e. beta) this indicates that PAR is not sufficiently high to constrain the shape of the LRC
set to TRUE to estimate nighttime respiration based on basal respiration estimated on nighttime data instead of basal respiration estimated from daytime data. This implements the modified daytime method from Keenan 2019 (doi:10.1038/s41559-019-0809-2)
list with entries of given arguments.
partGLControl(nBootUncertainty = 40L)
#> $LRCFitConvergenceTolerance
#> [1] 0.001
#>
#> $nLRCFitConvergenceTolerance
#> [1] 0.001
#>
#> $nBootUncertainty
#> [1] 40
#>
#> $minNRecInDayWindow
#> [1] 10
#>
#> $isAssociateParmsToMeanOfValids
#> [1] TRUE
#>
#> $isLasslopPriorsApplied
#> [1] TRUE
#>
#> $isUsingLasslopQualityConstraints
#> [1] FALSE
#>
#> $isSdPredComputed
#> [1] TRUE
#>
#> $isFilterMeteoQualityFlag
#> [1] FALSE
#>
#> $isBoundLowerNEEUncertainty
#> [1] TRUE
#>
#> $fixedTRefAtNightTime
#> [1] NA
#>
#> $isExtendTRefWindow
#> [1] TRUE
#>
#> $smoothTempSensEstimateAcrossTime
#> [1] TRUE
#>
#> $isNeglectPotRadForNight
#> [1] FALSE
#>
#> $isNeglectVPDEffect
#> [1] FALSE
#>
#> $isRefitMissingVPDWithNeglectVPDEffect
#> [1] TRUE
#>
#> $fixedTempSens
#> E0 sdE0 RRef
#> 1 NA NA NA
#>
#> $replaceMissingSdNEEParms
#> perc minSd
#> 0.2 0.7
#>
#> $neglectNEEUncertaintyOnMissing
#> [1] FALSE
#>
#> $minPropSaturation
#> [1] NA
#>
#> $useNightimeBasalRespiration
#> [1] FALSE
#>