|
<br />2-01
<br />
<br />SEX:TION 2 .., GEN)!:RAL ~~ABILI',rI CONCEPTS',~ Dl!;1"JJU'J:.rq~
<br />,-.._"., ".. l_'" 3 ....., ',...,
<br />.~
<br />, .-.,..
<br />2-oL . INTROD1.lCTIQN , .
<br />
<br />, - ~, '" -' ,,' "- '-... -. -" . '\ - .", . .-. ' .' ,
<br />The subjects .ot probabilitY e,Da... stAtil!ticsare beeQlll1~.~
<br />increll.8ingly applicab1e'in erig1nee.rtng work, arid. it 1s -con;". .,..:
<br />8idered apPropriate toprov{de pat1~unil.1n:r()~t;t9l1..~ ,:,,:~
<br />orlenting those engineel'l! who b,e.~not ~ torma.l. tra1~g"l~.-, <-
<br />the subjects. This dectloncontairi!l .. baa!. ~eW. 'ot ,j;~,::','~'~
<br />theoretical basis. tor. i?iobabi1i ty 'est1lDates paSl;td. 'On .oDs,e",~';~ ::~_
<br />data. Details ot aPl;llication Will' be :Presented 'later,aild' II. ..'
<br />b:roader understanding at the theory can be obtained t:rom text-
<br />books such ll.8 re:1'erence 1.
<br />
<br />2-02. NA!l'URE OF RANDOM JSY~
<br />
<br />!!,. Probability estimates made in hydrologic engLneering
<br />e.re based on records ot random events. To understand p:robabil1 ty
<br />methods and tully appreciate the degree ot reliability 0:1' such
<br />probabili ty estimates, one should consider the Illl.ture and varia-
<br />tion 0:1' random samples.
<br />
<br />E.. Consider II. period 0:1' 2,000 years during which con-
<br />t:roll1ng hydrologic conditions do not chs.nge. Annual maximum
<br />hydrologic events occurring during this period can be divided
<br />into 100 records ot 20 years each. From knowledge 0:1' probability,
<br />it is expected that one 0:1' these records will contain a :1'lood
<br />that is exceeded on. the average only once in 2,000 years, a very
<br />re.re event. About 18 ot these records will contain f'loods that
<br />e.re exceeded on the average only once in 100 years (it youJ.d be
<br />20, except that some of' the records might contain more than one
<br />, h,,1.lld ""ll ~aJ.n
<br />:1'loods larger than that exceeded on the average once in 20 years.
<br />On the other hand, ll.bout 12 ot the records would not have t100ds
<br />larger than that exceeded on the average once in 10 years.
<br />
<br />c . When II. hydrologic engineer is studying a record ot 20
<br />years'-length, he cannot tell by examining the record alone whether
<br />it is one that has a norma.l sequence 0:1' events, abno~ rare .
<br />events, or ll.Il abnorma.lly slllll.ll number ot large events. It the
<br />record contains abnorma.l1y large events, the resulting p:robabil1ty
<br />estimates tor l.s.rge events Will be too high, and vice versll.. In
<br />order to reduce the uncertainties t:rom this source, it is advis-
<br />able to study &ll 0:1' the events in relation to each other and to
<br />introduce knowledge obtained on similar phenomellll. at other locations.
<br />
<br />-4 -
<br />
|