Laserfiche WebLink
<br />30 <br /> <br />during that period. It has since been closed. Obviously, Colorado's <br />high elevation precipitation measurements have left something to be <br />desired. For future research and analysis, we must work hard now to <br />establish and preserve high quality, year round precipitation stations <br />at fixed locations in the Colorado mountains. <br />C. Variability of Colorado precipitation <br />The 1951-80 precipitation map is a graphic visual demonstration of <br />the variation of annual precipitation in complex terrain. It shows only <br />the average precipitation and gives no information about how variable <br />precipitation is from one year to the next. Fortunately, some measures <br />of the year to year variability of precipitation are not nearly so <br />dependent on the terrain as precipitation itself. If precipitation was <br />normally distributed, then the preferred measure of variabil ity would be <br />the ratio of the standard deviation to the mean. Since precipitation is <br />not normally distributed, the cumulative distribution of the probability <br />of nonexceedance is a better indicator of variability. . <br />Cumulative distributions can be developed to obtain nonexceedance <br />probabilities both empirically and mathematically. The Gamma function <br />is well known for its ability to produce an accurate fit to an actual <br />distribution of precipitation data. The advantage of using the Gamma <br />function is that it smooths some of the inherant noise from a <br />distribution of real data and makes it easy to calculate the probability <br />of nonexceedance as a function of precipitation. Because of the <br />smoothing process, comparisons among a number of stations are less <br />affected by natural "noise" in the precipitation data, <br />An example of the cumulative distribution produced both empirically <br />and mathematically (employing the Gamma function fit) for Fort Collins, <br />