Laserfiche WebLink
<br />The generalized skew coefficient G is to be estimated from a group <br />of stations with 25 or more years of record surrounding the station of <br />interest. <br />Skew values from 40 stations or at least all stations within a <br />radius of 100 miles should be plotted on a map at the centroid of the <br />basin to determine if any geographic or topographic trends are present. <br />If a pattern is evident, isolines are drawn and the standard deviation <br />of the differences between observed and isoline values are used in <br />appraising the accuracy of the isoline plot. It may be possible to <br />improve values from the plot using prediction equations as described <br />below. If the standard error of estimate from the prediction equation <br />is smaller at the 5 percent level of significance than the standard <br />error of estimate from the isoline plot, the prediction equation should <br />be used to estimate the generalized skew coefficient. If the prediction <br />equation does not provide an improved estimate, the generalized skew <br />coefficient should be taken from the isoline plot. <br />The prediction equation should be based on data from at least 40 <br />stations with a minimum record length of about 25 years. The data <br />should include all stations within the same meteorologic, topographic, <br />and geologic region as the location of interest. A minimum radius of <br />100 miles is considered appropriate. It is recognized that in some <br />portions of the western United States a relaxation of these criteria is <br />necessary. <br />The prediction equation would relate either the skew coefficient or <br />the deviates from the isoline map to predictor variables that affect the <br />skew coefficient of the flood reco~d. These would include such variables <br />as the drainage area, basin shape, an index of channel or basin slope, <br />geographic location, distance and direction from the coast or from some <br />other significant topographic feature, and a meteorologic and/or climato- <br />logic variable (e.g., annual precipitation). <br />The prediction equation should preferably be used only for estimating <br />the skew coefficient for stations with variables that are within the <br />range of data used to calibrate the equation. <br />If a pattern is not evident in the isoline plot, a prediction <br />equation should be attempted; if this is not significant, then an <br /> <br />13 <br />