WMOD00511 - Laserfiche WebLink

Home Browse Search

...... . .._,~n<'~~l~;X;~~)~~'f~g~jji~~ 686 B, SEVRUK AND L. ZAHLAVOVA was adapted to a 'fine scale' during the test (Table I). This was found more convenient, primarily in cases where no clear division between the basic four classes and three interim classes seemed to be possible. In the field, the angle :x was measured by an optical instrument positioned immediately above the gauge orifice. Angle (1. was measured along the eight directions of the wind rose and averaged. Obstacles such as houses, bushes, and trees at a distance of no more than 300 m from the gauge were taken into consideration. Different heights of obstacles in a particular sector of wind direction were considered by taking one average value (1. for a sector. In cases of individual trees or other free-standing obstructions whose horizontal extent was less than one-tenth of their distance from the gauge, (1. was reduced by half. If:x values for one or more sectors were greater than 250, only the value of 25 0 was considered for the computation of the average of (1. for a particular gauge site. This was necessary because in some cases the values of (1. exceeded even 500 in one or more overprotected sectors, but were only up to 10-200 or even less in other sectors. Consequently, including such a large value would result in an unrealistic average value of (1. for a particular gauge site. In addition, a comparison was made for nine Swiss gauge sites between the average :x value as computed from 8 and 16 sectors of the wind rose. Furthermore, weighted averages of :x were also computed by multiplying by 2 the :x values for the westerly sectors, SW, W, and NW, from where the prevailing precipitation-bearing winds in Switzerland originate. The averages of rJ. for the westerly sectors were also computed and compared with the eight-sector average. RESULTS t The frequency and average values of differences between estimates and measured GSE classes for the three test people are shown in Figure 4. It indicates that the degree of experience can have a considerable effect on the accuracy of GSE estimates, but the average difference is, in general, rather small. It amounts to less than :t 0,66 of a class for an inexperienced student: :t 0,5 of a class for a less experienced young assistant, and approximately :to.33 of the class for an experienced, older assistant; amounting to less than :t40 of the vertical angle rJ. of obstacles. As can be seen from Table I, the exposure classes differ from one another by 7 v. Small differences, equal to or less than :to.5 of a class occurred in more than 80-90 per cent of events in the case of experienced assistants, particularly for the more experienced assistant. The inexperienced student made errors of greater magnitude, with differences of :to.75 of a class (50 per cent of events) up to 2 classes (1 event out of 50). Moreover, almost 20 per cent of all differences showed magnitudes between I and 2 classes. An interesting feature was that the student regularly underestimated protected gauge sites, that is classes 3 and 4, and overestimated less open gauge sites, classes I and 2. The trials showed that, regardless of the experience of the assessor, the accuracy of GSE class estimates depends primarily on the quality and completeness of the station history records. The best estimates are 40 F[%] ~ 30 20 10 0 0 1/2 1 1/2 2 0 1/2 1 1 1/2 o [class] Figure 4. Frequency. F. of differences. D. between the estimated and measured exposure classes for an inexperienced person (on the left) and two experienced analysts (on the right); the darker columns on the right indicate the more experienced analyst .,'.,.".....~"""'''~''''''''ll~-~c . ",~, ~ ,".'~ .,', -' ~ :''''', ..:t~~~~'f;t?.