Figure 4
From: Observations of enhanced rainfall variability in Kenya, East Africa
Wavelet power spectra of spatially averaged rainfall over arid agro-climatic zones of Kenya (a) and Narok (e). In (a) and (e), color contours represent wavelet power normalized by the squared standard deviation and are at 0, 1, 2, 3, 4 from the lightest to the darkest colors, solid black contours enclose areas of > 0.95 significance relative to a lag-1 red noise spectrum, hatched areas represent the cone of influence—area of the wavelet spectrum where edge effects (errors) due to a finite time series cannot be ignored. Figures (b), (c), (d) and (f), (g), (h), respectively show scale averaged power over the periods 18–30, 5–10, and 2–4 months for (a) and (e), and show average rainfall variance for those periods. Dotted horizontal lines show the > 0.95 (black) and > 0.99 (red) significance levels. Curves above the dotted lines have significant variance. The power spectra in (a) and (e) were generated using pyCWT101 Python library for spectral analysis. All the figures were created in Python 3.10.9 using Matplotlib 3.7.087 library (https://ieeexplore.ieee.org/document/4160265).
