Fig. 2: Relationship between wind power density (WPD) and temporal resolutions (∆t).
a Wind speed distribution for different ∆t values (1-hour to 48-hour) fitted with Weibull probability density function, using wind speed observational data (2014–2021) from a rural station in southern Canada (52.3°N, 111.8°W; same site as Supplementary Fig. 2). Parameters k and c describe the Weibull shape and scale, respectively, and the numbers in the panel following k and c (e.g., 1, 3, 6, …) indicate the corresponding ∆t values (e.g., 1-hour, 3-hour, 6-hour, …). WPD for this example was calculated based on the statistical method according to Eq. (5). b Exponential fits of WPD versus ∆t for four global sites representing different WPD levels: Site 1 (southeast Australia, 38.4°S, 141.6°E, WPD: ~400 W m-2), Site 2 (Kazakhstan, 49.6°N, 73.3°E, WPD: ~300 W m-2), Site 3 (southern Canada, 51.7°N, 112.7°W, WPD: ~200 W m-2), and Site 4 (southern USA, 35.4°N, 94.8°W, WPD: ~100 W m-2). Each site’s fitted curve is shown in a different color. Parameters a and b for each site are labeled as a1, a2, a3, a4 and b1, b2, b3, b4, respectively. c Ratio of WPD(∆t) to WPD(1-hour) as a function of ∆t, corresponding to the sites in panel (b). Each site’s fitted curve is color-matched with (b). d Scatter plot of WPD(1-hour) versus parameter a, with *** denoting a correlation exceeding the 99.9% significance level in t-test. e Scatter plot of WPD(1-hour) versus parameter b. f Scatter plot of parameter a versus b. g Global distribution of parameter a. h Global distribution of parameter b. i Global distribution of the coefficient of determination (R2) for the exponential fits of WPD versus ∆t. Sites with R2 < 0.7 (accounting for less than 4.2% of all sites), which show abnormal WPD values and negative b values likely due to data anomalies, were excluded from the analysis. The fitting was performed using 1-hour resolution wind speed data and Eq. (1), with parameters a and b determined for each site. Source data are provided as a Source Data file.
