Laserfiche WebLink
<br />\ J <br /> <br />.i 3.2. DISPLAY AND USE OF ESTIMATED FREQUENCY CURVE. The <br />estimated frequency distribution is displayed on a grid with the magnitude of the event <br />as the ordinate and probability of exceedance (or percent chance exceedance) as the <br />abscissa. The plot thus provides a useful tool for estimating quantiles or exceedance <br />probabilities. Specialized plotting grids are available for the display. These grids are <br />constructed with the abscissa scaled so a selected frequency distribution plots as a <br />straight line. For example, a specialized grid was developed by Hazen for the <br />commonly-used normal frequency distribution. <br /> <br />The specialized normal-probability grid is a useful tool for judging the <br />appropriateness of the normal distribution as a model of the parent population. If data <br />drawn from a normally-distributed parent population are assigned plotting positions <br />using equation 1 and are plotted on Hazen's grid, the points will fall approximately on <br />a straight line. If the points do not, then either the sample was drawn from a <br />population with a different distribution, or sampling variation yielded a non- <br />representative sample. <br /> <br />A specialized plotting grid has been developed also for another commonly-used <br />frequency distribution, the log-normal distribution. If the data truly are drawn from a <br />log-normally-distributed parent population, the points will fall on a straight line. <br /> <br />4. ANALYTICAL TECHNIOUES. <br /> <br />Analytical techniques define the relationship between streamflow <br />magnitude and probability with analytical tools, instead of the graphical tools. With <br />numerical techniques, the following general steps are used to derive a frequency curve <br />to represent the population (McCuen and Snyder, 1986): <br /> <br />. Select a candidate frequency model of the parent population. Three <br />distributions are commonly used for frequency analysis of hydrometeorological <br />data: the normal distribution, the log-normal distribution, and the log-Pearson <br />type III distribution. <br /> <br />. Obtain a sample. <br /> <br />. Use the sample to estimate the parameters of the model identified in step 1. <br /> <br />. With the model and the parameters, estimate quantiles to construct the <br />frequency curve that represents the parent population. <br /> <br />4.1. NUMERICAL PARAMETER ESTIMATION. Parameters of a statistical <br />model commonly are estimated from a sample with method-of-moments estimators. <br />The method-of-moments parameter estimators are developed from the following <br /> <br />7-95 <br />