Home

Welcome to the loadflex wiki!

Peer-reviewed documentation: Appling, Leon, & McDowell (2015)

We have described this package and demonstrated its use in Appling, A. P., M. C. Leon, and W. H. McDowell. 2015. Reducing bias and quantifying uncertainty in watershed flux estimates: the R package loadflex. Ecosphere 6(12):269. http://dx.doi.org/10.1890/ES14-00517.1.

The paper is open access - please let us know what you think! Our favorite mode of communication is via GitHub Issues (https://github.com/McDowellLab/loadflex/issues), but email also works fine.

Corrections and clarifications to Appling, Leon, & McDowell (2015)

• In the equation on page 10, sigma_p should be squared.

• On page 10, we wrote: "the residual-corrected predictions are either retransformed with the same method as for linear regressions if the residuals correction is done in log space, or left in linear space if the residuals correction is done in that space." That's not quite right. Predictions from loadLm or loadReg2 (the regression models) are bias corrected and returned to linear space before being passed to loadComp. When loadComp (the composite correction) is applied with log space selected, predictions are logged, corrected for residuals, and then exponentiated without any further bias correction. This ensures that the final composite method predictions run directly through the original observations in most cases. ("Most cases" includes composite corrections using rectangular, piecewise linear, and spline interpolations. Smooth spline interpolations don't and shouldn't have this property.)