This function provides a test to decide whether low-frequency modulations in the relationship between climate and tree-growth are significantly stronger or weaker than could be expected by chance.
g_test(x, boot = FALSE, sb = TRUE, check_duration = TRUE)a data.frame with p values for the testing the null
hypothesis that the low-frequency modulation of the
correlations of the variables with tree-growth can be
considered as noise.
an object of class '"tc_dcc"' as returned from a call to
dcc with moving correlations enabled
logical shall the individual correlations be
bootstrapped? (see details)
logical shall a status bar be drawn?
logical should the duration be checked before
running with individually boostrapped correlations? The default is `TRUE`,
set to `FALSE` to suppress interactive selections and messages to the
console.
This function is a multivariate extension of the test for spurious low-frequency modulations for moving correlations of time series as proposed by Gershunov et al. (2001). In short, 1000 simulations of random data sets are generated, where the climate data is simulated as Gaussian noise, and the tree-data as linear combinations of the climate parameters using the original coefficients of the correlation function, and an error component with a variance equal to the variance unexplained by the individual parameters.
For each iteration, a moving correlation function is calculated with exactly the same settings as the original model. The standard deviation over the individual windows for each parameter is then compared to the bootstrapped distribution of the standard deviation of the simulated data to test for significantly higher or lower low-frequency modulations.
Gershunov, A., N. Schneider, and T. Barnett. 2001. Low-frequency modulation of the ENSO-Indian Monsoon rainfall relationship: Signal or noise? Journal of Climate 14:2486-2492.
# \donttest{
dc_cor <- dcc(muc_spruce, muc_clim, 3:9, method = "cor", moving = TRUE)
g_test(dc_cor)
# }
Run the code above in your browser using DataLab