It is okay to be average when quantifying rangeland dynamics Comment on: Easdale, M.H & Bruzzone, O. 2015: Anchored in 'average thinking' in studies of arid rangeland dynamics e The need for a step forward from traditional measures of variability. J. Arid. Environ. 116: 77-81

Abstract

Easdale and Bruzzone (2015) recently indicated that to better capture temporal patterns of rangeland dynamics, there is a need to move forward from simple measures of variability such as mean, standard deviation, and coefficient of variation (CV). They note that this is especially true in cases where long time series are available (e.g., ten or more years at a monthly resolution), as with remotely sensed data. To support this viewpoint, Easdale and Bruzzone (2015) presented multiple simulated time series datasets that had same mean and standard deviation, but different seasonal patterns. These seasonal pattern differences were only captured using Fourier analysis (i.e., power spectrum analysis, Chatfield, 1996), highlighting the utility of the method. Here, however, we show the opposite phenomenon, where multiple datasets can show similar temporal patterns from Fourier analysis, but have different and meaningful means, standard deviations, and CVs.