The Frequency Distribution of Lagos/Ikeja Wind Gusts

View More View Less
  • 1 Nigerian Meteorological Services, Lagos
© Get Permissions
Restricted access

Abstract

In a majority of cases, the frequency distribution of maximum values of meteorological elements is either normal or skewed positively, although good agreement with the normal distribution is not generally found. Several types of extreme-value distributions are available. Gumbel's 1954 distribution, the most widely tested and used, appears as a straight line when plotted on extreme-value probability paper and extrapolation is simple.

Daily maxima of homogenous wind data for Lagos (Ikeja) (06°35′N, 03°20′E) for 1948–1962 have provided 4033 observations. These were first plotted as an actual frequency distribution, smoothed, and then plotted on extreme-value probability paper. The reduced variate y is related to the frequency p by the relation y = −ln(−lnp). The straight line Gumbel distribution, the equation of which is x = u + (1/a)y, where u is the mode and 1/a, the slope of the line, is related to the standard deviation of the sample population. Since these data fell along a straight line, Gumbel's distribution provided a good fit. The larger the number of observations the more closely Gumbel's theory tends to apply.

Abstract

In a majority of cases, the frequency distribution of maximum values of meteorological elements is either normal or skewed positively, although good agreement with the normal distribution is not generally found. Several types of extreme-value distributions are available. Gumbel's 1954 distribution, the most widely tested and used, appears as a straight line when plotted on extreme-value probability paper and extrapolation is simple.

Daily maxima of homogenous wind data for Lagos (Ikeja) (06°35′N, 03°20′E) for 1948–1962 have provided 4033 observations. These were first plotted as an actual frequency distribution, smoothed, and then plotted on extreme-value probability paper. The reduced variate y is related to the frequency p by the relation y = −ln(−lnp). The straight line Gumbel distribution, the equation of which is x = u + (1/a)y, where u is the mode and 1/a, the slope of the line, is related to the standard deviation of the sample population. Since these data fell along a straight line, Gumbel's distribution provided a good fit. The larger the number of observations the more closely Gumbel's theory tends to apply.

Save