search

lawinslow / AncientIce

1 stars 0 forks source link

Comment 13: for luke, missing data and spectra #24

Open rBatt opened 9 years ago

rBatt commented 9 years ago
  1. Comment: The wavelet spectra (Figure 3 panels C-D) appear to span the entire period, with just a single influence cone, despite there being gaps in the records. I could not understand from the Methods section how these gaps were handled in computing the wavelet spectra.

Luke says to put in constant value; this dampens variability.

Unsure of other ideas.

rBatt commented 9 years ago

Related to #28

rBatt commented 8 years ago

@lawinslow did you address this? I remember it coming up in emails

lawinslow commented 8 years ago

Torn is complete. The large gaps are no longer analyzed in the new analysis. Small gaps are filled with the mean of the timeseries, which tends to make the analysis more conservative and affects the short-scale variability over the longer-scale, which we're most interested in.