CLIMATIC NORMALS AS PREDICTORS. PART 3. MEDIAN VS. MEAN
SAN FERNANDO VALLEY STATE COLL NORTHRIDGE CA
Pagination or Media Count:
Rainfall and other variables with similarly skewed distributions are hard to characterize climatically due to their extreme variability. The arithmetic mean, generally used, is greatly influenced by extreme values. For rainfall data from 219 stations located in the western United States, the median was found to be a more representative value, and somewhat better than the mean for predicting future rainfall amounts. Some monthly precipitation frequency distributions are so greatly skewed that values smaller than the mean occur 90 of the time. Because any single measure of central tendency is inconclusive, measures of absolute and relative variability are summarized. Maps of percentage occurrence of the mean, ratio between median and mean, coefficient of variation CV, and relative variability Vq are presented for the mid-season months--Jan, Apr, Jul, and Oct.