Вы находитесь на странице: 1из 19

4- 1. ERRORS.

a. ~
n.11 e rr o r i s detined a s the differen ce between the ' .true
valu .: ct n r.l t he measured value of a quaot i.:-t y. ~tis a devia-
tion of an observat ion or a calculat ion fxom the true
va l ue and is often beyond the - c~ntrol of the on~ perfor~in g
the operatio n. Since the true value of a quantity can never
be asc e rt aine d by measurem ents, the'. exact va.lue of an
error , li ke~ise , can never b~ determin ed in any measure-
ment. . ,
Errors are inherent in all measurem ents and result from
s our c e s wh i c h cannot be avoided. They'may be caused by the
type of equipmen t used or by the way in which the ~guipmen t,
is employed . I t may a~so be caused by the impe~fec tions of
t he senses of the person undertak ing the measurem ent or by·
natural causes. T~e effects of · errors cannot be entirely
eliminate d; they can , however, be minimize d by - careful work
and by applying corre c t ions.
In ~ny s urve yi ng n j eration the surveyor is continuo usli
dealing wt th errors. I f the work must be performe d to
exacting sta ndards he mu s t understan d thorough ly the dif-
ferent kinds o~ error, . their sources and behavio·r , magni-
tude, and effects upon field measurem ents . It is only by
then that he can intellige ntly select the instrume nts to be
used and the survey methods t9 be employed which wilJ
redu ce errors to acceptab le limits.

4-2 . MISTAKES .

Mistak~s are inaccura cies ~n measurem etits which occur


because some aspect of a surveyin g operation is performe d
by the surveyor with careless ness, inattenti on, poor judg-
ment, and improper e xecution ·. Mistakes are also caused by a
misunder standing . pf ~he problem, inexperie nce, ot indif-
ference of the surveyor . A .large mistake - is referred to as
a blunder. 'Mistakes and blunders are not classifie d as
errdrs pecause they usually are so large in magnitud e when
compared to errors. . -- .
Among students of surveyin g, mistakes which are fre-
quently committed include: reading the wrong graduatio n on
the tape, omitting a whole length of tape, transpos ition of
figure s , reading a scale backward , misplacin g a decimal
point, incorrec t recording of field notes, adding a row or
column of numbers incorrec tly, etc .. For exc\.mple, a tapeman
may re.ad a number on the tape as 6 when it should actually
be 9-, or he may read a taped dis ta nee as 4 8. 6 rn but records
i t as 46. 8 m in · the fiel d notes. Another example of a mis-
take is i'r, tt\e recordi ng of a series of repeated measure-
ments of a line. The tape man may record the measurem ents as
follo ws: . '241.85 m, 243.8 8 m, 234.80 m, 243.86 m, and 243.85
"'· If a- curs·ory inspecti on is made of the recorded values ,
lt will be noted fhat the third value disagree s signifi-
~~ntly with th~ o t hers . This apparent ly is because of
4-1 . ERROR S.
• •

An err or is de t i ned as · t h e d itf e •r ~·n·c e -~e t we~·n t h e t rue


v al u ,:.: and t he meas ured. va -l u.e df a · qu~·o ti:ty . ~t ·.)s. a dev ia -
t i ori of a n obs :ervat i on qr a . C:a.l ~ul a t _i oh _; ~i;o'~ . tpe \ ·:true
va l ue and i s o ft e n peyond the - cont t- o l' .o f t he or,i~ · ~e r f o.riqi n g
the oper a tion . Sinc e t he . true ··v a lue o.f a . quant i ty .ca n n eve r
be - a sce ~ta 1.ned . by . me~ s u r e me ri t s, . t 't.1.e' , exa c t -va .lue · o f an ·
er·ro r , 1 ike°\" iSe , c~n n e v e r be d e t erm i n ed dn any 'me a s ure -
'. . . ,
me n t. ,. . . .
'
.
Errors a re inhe rent i n a l l .meas u re ments a nd · r es ult f r om ·
so ur ces wh i ch c a nnot be a voide d·. They' inay b e ·ca'.,µ s ·~ d qy · the
type of e q u_ipme_n t used or by th e way : in . wtd:ch . t-h e · eCJu ipme nt ,
_is e mpl oye d . I t ma y a ~so be c aus~d by t;he i mp·e;r: emen. f -e ctions o f
the se n s e s o f the p er s on u nd e r tctk in g, t he mea s u r t or by ·
natura l c ause.s . 7
1.'h e effec ts o -f ; err o rs c ann o t be . en tirely
elimin ate d ; ~~ e y ca n, ho wever, tie mihi~ ized b y r o a re f ul wo rk
and by ~pp l~in g correc t io ns . · · ·
In a ny survey i n g 0 0erjti on th e ~ui~e i~r is -~ o n tin uo~s l y
deal i ng wi t h . e rro rs . I f t h e wo rk must ·be . per forme d . to
exacti ng s t a nd a ~ds he mu s t und~rs tarid .t h~ ro u~hly t he di£~
fe r e n t ki n d s of e r ror , t h e ir source s a nd . be h a v i or, . . .ma g ni-
tude , and . e fiects up~ n fi~ld measu re~ent s . It i s only b y ·
t h e n that h e ca n i n te llige n tly selec t t h e i n str ume nts -t o be
u sed a nd th·e survey method s t9 be· e mpl oye d wh-i-ch wilJ
red u ce er rors t o accep ta b l e l im i ts .

4-2 . MI STAKES.

Mi s t a k'es are inaccu rci'cie s i, n · measu remen ts _ whic h occur


becaus e' some aspec t of a sur v ey i ng opera tl on is -_· p e rf o rmed
by t h e · s u~ v eyo r ~it h carele ssness , i natte~ tioni prior judg~
ment , a nd improp er e ~ecuti on ·. Mi stakes are a l so '. c a u s ed by a
misund ersta nding . pf .th e pro bl em , . i nexpe rience , or ·_ ,indi f-
f e ience- of t h ~ survey o r . A ' la r ge mistak e - is referr ed to as
a blund e r. Mistak es and blund ers are not c la~sif ied a s
. . . --
errdr s heca ~se they usua ll y are so large .in · mag n i. tude
·
. . . -
when
compa red to errors .
Among studen ts of survey i ng, mistak es wh ich are fre -
quentl y c ommi tted in clude : readin g the wron g gr a dua tion on
th~ tape, omi tting a whole length of tape; tra nspos it i on of ·
figure s, · re ad i ng a scale backwa-rd, mispla cing a decim al
point , incor~ ect recor~ in g of fi~ld n otes , a dding a row or
column of numbe rs in co r rectly , e t c .. Fo r ex~ mple -; _a tapema n
may re.ad a numbe r on th e ta p e as 6 wh e n .it s hou~d actua lly
be_ 9·, or he ma y r ea d a t a p e d dista nce a s 48. 6 rri. but record s
it as 46.8 rn in · the fi~ld riotes . _.· Anothe r examp le of · a mis-
t a ke is ir, tne rec-ording of a series. of repeat ed measu re-
me nts of a line. The t apeman may record th e measu remen ts as
· f- o 11 o ws : . 2 4 J • 8 5 m" 2 4 3 • 8 8 m, 2 3 4 . 8 0 m, 2 4 3 • 8 6 · m, and 2 4 3 • 8 5
.
tr\. If. a., cursor y inspec tion is. made of the record ed values , '

w1l L . be noted fhat the third value disa g rees signif i -


,

it
cantly with _th~ ~thers . This appar entl y is becau se of
tra ns po s ition of fi g ur es in th~ process of r e cord i ng
If c a r ef ul ·a ttentio n i s gi~ n to the ex e cution ~ f a
. t ' . ny
5 ur v ey1~g op.e ,ra. io n, mista kes c an be a voided . Th e various
types o f mi s tak e s , howeve r : c a n only .b e corr e cted if di s c o-
ver ed. The i r de~ect i on c ~n _b e made by s ystema t ic check ing
of a l l wo r ~, making a cvmmon sense es t i mate and a na l ysis ,
o r b y maki n g a d u p licate me a s ~r ement .
Unde·te c t~d -, mi s t akes ma y p r oduce ve r y ,serio us e f fe c ts on
t he final result of A s urvey such ·that the s u r ~eyor sh o ul d 1

e x e r t • eve r y p oss ibl e e ffort t o minimiz e thei i oc c urrence .


When mi s .t a k es are dis c over e.d ,_ c orz;eoti oris sho ul d immedi at e-
ly bf made. I ~ i s usually advisa ble t 6· re pe at a mea s uremen t
when a mis~ ak~ i s, det ected . Howeve r, if a n adequa t e ~umbe r
of othe r rnia s u r eme nts of th ~ s~fue q uanti t y are · avai l abl e
and which c losel y · ag re e , · th e wide ly dive rg e n t r e s ul t s
should be di ~c a rded . Mistake s have •p6 place iri ' e ngine e r i~ g
a nd su r v e y i n g wor ks . The effects of. er ro rs can be minimi zed,
but t hey ~a nn ot be e n ti r ely elirnina t~d , wher e as mistake s
1
c a ~ be d e t ect e d a nd 'remove d fro m reco r de d me as ureme nts ~ .

4- 3 . TYPES OF ERRORS .

I n s ~ r v~y i ng , errors in measure ~~nts are ei th er s y ste ma ti c


or a cci de n tal . These are defined in terms o f the i .r beha v io r
as to sign a nd fuag~itu de, and the conditi ons pre v a i l i n g
during th ~ measure ment .

1. System ati c Errors . Thi s t y pe · of error is one whi ch


wi l l always ha ve the Siame sign and magnitu d~ as long a s
f i eld condi t ions remain c onstant and unc ha n ged . For c ha n -
· g i n g fie l d cond'i t i o ns ' there . il a corresp onding _change in
ma gn itude of the error', ho wever I the sign remains consta nt.
A system atic er r or will repeat itseJ.f in other measur e-
ments, still mainta ining the same sign, and thus will ac-
cu~ ulate . It is for th i s reason that this type 0£ error i s
als o cal~ed a cumula tive erro, . For instanc e., in making a
measure min,tw ith a 3_0-mtap e.which is 5 cmtoO short , · t h e
same error is made.ea ch time the tape is used. It a ' full
t ape ' length is u::, ed six times , the e.:,;ror accumu lates a nd
t ota t ~ si~ time s the~eri or (or 30 ~m) ~or the total mea-
s u rement .
'syste mat i c errors conform to. mathem atical and phys i ca l
1 a ws . s u c.h err qr s can be com~ u t e d and the i r e f f e ct s e 1 i mi -
nated b y applyin g correct ions , employ ing proper tec hni q u e s
- in the use of . instrum ents, or by adoptin g a field pr o ced ur e
wh ic·h ·wi ll automa tically elimin? te it .
r-n s ur veying , system atic errors occur due to · i n strume n-
t a l t~ctors . - natura l cactses, and human limitat io n s of th e
obs erve r. Th i s type of error will co n tinue t o pe r s i s t and
impose resu la r ~ffects in the perform a n ce of a s u r v e y ope-
r atio n.
Accide nta l
__ 2-,., Error s . Thes e er r o rs a re p ur e ly ac -
c ident a 1 • in char a cter . Th e ~ ccur r e nce of s~ ch e r rors a ~e
/ NTR OOUC'riON
TO SURVE r tNO
-----~33 ,
matt ers o f cha ncqa ius they ~ re li'ke ly to be posit -ive or ne -
1

gat f ve, and may tend ln paz:t to co~p en~a te· or avera
ge out
acco rding to· laws Qf prob abili ty . Ther e is no abso
lut e wa y
of deter mini ng oi elim inati ng them si6c e the erro r
for an
obse rwat ion of a quan tity is not like l,t. to be the same
· as
for a seco n~ - obse rva tiqn ·. ' ·
Acci dent al erro r s are c ause d by facto rs ' beyo nd
the
c ontr ol of . the. surv eyor and are pres ent in all .sur
v ey ing
meas ur emen ts . Thl! y rema in ,afte r mt.s take s and syste
mati c
erro rs . have bee-n· elim i nat~d . An exam ple of such an
er r o r is
the failu re of the t~pe lJlcln to exer t the corr ect amou nt o f
pu l l on the ends of ·a tape durin g meas urem ent . Some
times he
may exer t a pull beyo nd that whic h .is r ·equi red, and s ome -
time s less ;:han what ts r e gu_i red. Anot her ~xam pl e ia
-in the
read~ ng, of an angl e with a tran sit . Sinc e the instr ume
ntman
can n ot - read i t p er fece ly , there woulQ be time s whe n he
.woul d . read ~ va l u e whic h is too lar ge and i n anot her
·atte mpt h e .may re a d a v a l ue whi_ch woul d b e too smal
l. Thes e
err ~s ar e li kely t o have eithe r a pos i tt v e o r a
ne gativ e
s ign, a nd tend ·t o canc el eac h o ther or comp ensat e
for eac h
othe,r . • ·
In comp a ri so n t o syste mati c erro rs, accid e n tal e rrors
are u sua lly o f mino r impo rtanc e in sbr v e yin g ope
r a tion s
pi n c e t h.~ y are varia ble in sign and are of a .. comp
e n sat ing
natu re . Thi s char acte risti c usua lly tend s . to balan ce
o u t in
the fina l resu lts. Altho ugh the tot~ i erro r i n crea ses a s
th e n umbe r o f meas urere nts i ncre ases , the tota l erro r
come s prop ortio nally less when comp ared with the numb be-
e r of
meas u reme nts , and the accu racy becom e~ grea ter as t h e num-
he r o f meas urem ents ·incr ease s.

~ •· -4 . SOUR CES OF ERRORS -.

Th e s ou rces of erro rs in sur veyi n g ~eas urem ents are


c l ass i -
f i ed ~ nder the follo wing three grou ps .

1 . ·Inst rume ntal Erro rs e Th ese error s4' are due t o impe
fect i ons : n the instr um~ nts used , r-
eithe r from fau l ts in
thei r cons tr'uc t i .on or fr om i mpro per adju stme nts betw
ee n th e
d~ff ere n t pa~t s prio r to thei r use . Surv eying i nstiu
me n ts
jus t li ke any othe r in~tr umen t, are n ever perf ect ; prope r
c ~rre ct ton s ·and field meth ods are app l ied to bring
th~ mea-
s u re me n ts with in certa in allow abie lim i ts of prec
is ion .
Mo re ove r , with t i me a~u conti nuou s u sage , t he wear
a nd tear
o f th e i~st r umen t wil l like ly -be a cause for erro
rs . Exa m-
pl ~s of in strum e n tal erro rs are :

a ) _M eas tirin ~ wi t h a stee l t ~ ~e of i nco~ r ect l e ngth


.
b) Us ing a le v el in g ~oa with pa int ed g~ad uat ion s
not
p?rf e c t ly ~ paced .
c) Dete rmin ing t h e d iffer ence in ele v a tion b e tween
two poin t s with a n in s trum e nt whr se l i ne o f s ight
is not iri
a 9 j us t me n t . · ~

.
l ,VTROO UCTI ON
, TO SURVEYIN 8
d ) Sighting r 1 a rod which is warped .
e) Improper adj~ s tment of the pl a te · bubbles of
:..r ans i ~-,. or l eve 1 .

2. Natural Errors . The s e errors are caused by varia -


t i ons in t _h e phenomena of nature su~h as changes in 111agne -
tiC d e ~lination, te mpe rature , humid ity, wind , ref~action t
g ravity, and curvatur e o f t he ear t h . Natural error s &L~
beyond • the con tr ol • o. 'f • ma n . Ho wev er I in ord er to keep the
re s ult i ng error s w1 Ln 1 n all owa b l e li miL s , necessary preca u-
ti o n s can b e take n. Met h0d s can als o be a do pt ed to su it
prevail i ng condi t i o n s :
Th e s urv eyor may n ot be ab l e t o tot a l l y remov e the
ca use of s uch e r rors but h e can min im iz e their e f fects by
making_ proper c o r rect i o n s of the r es u lts a nd using · good
judgme , n t . Commo n examples are:

a ) Th e effect of temperature v~riatfon on the l e ngth


of a steel tape ~ · .
b ) Error in the readings of the maqnetic n eed 1y{ du e .
to v ar i ations in magnetic declination .
c) Deflection of the line of sight due to .the effec t
,f the earth ' s cu r v a ture and atmospheric refract i on.
d) Error i n the measureme~t of a l ·ine _wit h a tape
bei ng blown sidewise by a strong wind.
e) Error in t~e me a s u r eme n t of a horizontal dista nc e
d ue tQ ~lope .or uneve n gro u n d .
-
3 . Personal Err ors . Th es e erro rs ari s e p rincipally f r o m
1_i mi t: at i on s · o f t he s ens es o f s i .J h t , touch a nd hear i n g of
the hu~an observ er whi ch a re l i ke ly to be erroneous or in -
a cc u .r a t e . Th i s t yp e o f £ a 11 i b i 1 i t y d i f f e r s f r o -i' o n e i n d i v i -
dual to anothe r an d ma y va ry du~ t o certai n c 1 ~~ums tance s
existing dur ing a me a su r e ment. So me p e rson~ l erro r s a re
constant, s ome are compen~mtiny, while others mAy b~ e rra-
tic. Perso na l errors are s ·rgn i'fican tl j' re d uced o r elimina-
ted as skil ls are develop e d in s u rve y ing operatio~s t hr o ugh
consta n t practice and experien c e. Er rors o f t h is type - ar0
also ~liminated b y e mp loyi ng a p p ro pri a te checki n g o f p rocL
dures in the taki n g a nd rec ordi n g o f me a s u rements. - 1 ypic a _
of these e r rors a r e :

a) Error 1n dete r mi nir. g a r ead i n g on a ro d wh ic h i s


o u t of pl u mb during sighti n g .
b) Error in ' the rneasurem~nt of a vertical c. n. !J ~f- whe n
th e c r oss hairs of the telescope are not positic n et! c or -
r ec tly o n the targ~t.
c ) Making an erroneous estimate of the reguL~ed p~ l l
t o be ap pli ed on a steel tape during measurement .

4- 5. ACCURACY AND P_RECISION.

Accur a cy a nd precis.:.on are t •.,ro =•~rms which- are con s tantl y

1Nr11oo u crt,w
ro SU/tVE Y/!i/1 &UU f. .
used in surveying , ho weve r , . their c.orrect meanings il t~
often mi s unde r 6t ood . Whil e accura c y may be syn o nymous ~itt
Preci s ion the two should not be used in t erchangeabl y. TL ·
I i t I !~
s urveyo~ s hould al wa ys a ttempt to obta n meas uremen s ~hict,
are not on l y accurate but a lso prec i se . '
Accur a cy indica tes how clos e a giver\ mea surement 1s tc
t he absolute or true va lue of t he qua ntity mea s ured. I t i~,
plies th~ clo se ness bet we e,n re la~ ed , meas ure ments and t helt
exp~ctations. The d1ffer e n~e be twe en the mea sured valu~ Of
a quantity a nd i ts actua l valu~ repre se n ts the tota l errot
in the measurement. As the me as •ured . value approach e s th e
a ctual value, the magnitude of the error becomes smallet
and smaller; q nd as the magn i tude of the tot a l er ror is de~
crea se d, the accuracy o f the measu r e me nt incr eases. There,
fore, a measurement is t erme d l ess a cc u rate i f ' it deviates
by a signi f icant amount f rom its e xpected va l ue, and it is
mo4e accura t e if the de vi ati on i s rela t ively s ma ll.
The following exa mp l e illust ra t e s th e mea ning of accu-
racy. A. lin e k no wn o r ac c epted to b e 1 00 . 0 00 m long is mea-
sur ed tw i c e with a s t e el t a pe . The f i r s t meas ured va lue is
100.003 rn a nd t he s ec ond i s ?9 . 9 95 m. The first me as urement
is s a i d t o be more a cc ur a te t han t he se cond meas ur ement
sinc e th e e r ror in t he me~surement i s on ly 0 .0 0 3 mas com-
par e d t o t hat of the se cond me asuremen t wh i ch i s 0 . 005 m.
Prec i s i on r e fers t o th e de g ree of ref i neme n t a nd con-
siste nc y with whi ch a ny phys ica l mea s ureme nt i s made . It is
~or trayed by t he cl o seness t o o ne a n other o f a s et of
~epe~ t~ d me ~s u~e rnents of a quantit y. Th us , if a s e t of ob-
~ervat~ons 1s c l o s ~ly clust ere d t ogethe r, the ob s e rva t i on
1s ~a~d t? have been o btain ed wit h h igh p reci s i o n. S incP
pr~c 1s 1 0~ r e lat es to the exp e rt ness o f man ipu l ati on on th;
part o f t he observe r or . to th e capabi l it ies of th e in s t ,
ment ·
u se d ' . it. requires th ru-
. e u s e of pre c ise in st ru me nt s under
ideal co ndit ions e mp loying th e b est techn i q ues.
A method f r eq uentl ~ ~se d to defin e an d disti ngui s h be t -
w~e n a c cu r acy and prec i sio n i s illustr a t ed in F i g 4-1 Th
!~i~~::ntgi;en ~~ o w possi?l e groupings of r i fle shots ·upo~
. a r ge _s . In Fig. 4-.la , th e r es u lts h
ver y precise be ca use t he ri fl e h t . . s own we re
t o e a ch other . Ho we ve r s o s wer e cl u ste r ed c l tise
s h o ts were located s ome ' di;~ey we~ e n ot accu rate si nc e the
the n apparent that me asurcm=~~= ~om th e b u~ l ' s e y e . I t is
c e s s aril y accu r a te i f th Y be preci se b u t not ne-
ey a r e clos e ly g~ d t -
abou t a value t hat is a·
1
tf ~oup e oge t h e r but
signi f i ca nt amouri t . . ~r e nt fr om th~ expe c ta tion by a
In F i g . 4-lb, the riflema f.
g et a s t~e r if l e shots were ~ ir ed acc~rately o n t h e tar-
bull' s e y e . The shots h Paced relac1 ve ly close t o the
w 1· , owe v e r were t .
e r e s i g ht l'y scat t ered with , no pre cis e a s the y
Thi s ill us t ra t ion i s used~~ e xres~ect to t h e bull ' s e ye.
be acc ura t e bu t no l p r ec i s e i~ l;~ n t hat meas ur e me nt s may
a b out th e expecte d va·1,, e bu t e ~ ar e we ll di s t ributed
+::r om h ... , ar e s1 ·t · · ·
~ ac other. Th e r e s u ·1 t h ~n1 i ca nt ly di s pe rs ed
s so wn 1n F 1.
- ' g· 4- l c portray
(a l (bl
GOOO PRECISI ON BUT POOR ACCURACY GOOO AC CURACY BUT POOR PRECISION
(Avera ge hit owoy from bull ' • eye! (Average hit concentr ated neor bull'• •Y•l

le I (d i
aooo PRECIS I ON AND 0000 ACCURACY POOR PRECISIO N ANO POOR AC CURAC Y
· (Average h it concen t rated within th• bull', •Y•l (No overage hit in bull ' • ' ey e )

accu -
r i fle shot s upon a targ et whic h are both prec ise a n d
v er y
rate as they were plac; ed wi tbin t h e bu l l 's eye and· 1

ca n show that
c l osel y clus tered . Using this il l ustra tion ~e
t h ey a r e
meas ::. .: ,:::~1en ts could be both p r ecise and accu rate i f
clos el y group ed . aroun d the expe cted valu e. The res u l t s
shown in Fig . 4-ld port ray rif le s h ots whic h are n o t a~~u -
d a ll
rate and not prec ise s in ce th e y were ~ i dely · scat tere
clus ter .:.
over the targ et frame . Th e r e is no evid ent sig n 1 of ·
ing and nearn esF of the r i fle sho t s to the bu1i s eyL.
· mad e
· I t . is desi rable for s u r ve y ing meas urem en t s to be
inc,re ase. i_n
wit h high prec ision . Un fo rt u nat e ly, howe ver , an
nate in-
• pre c ision usua lly • wa rran ts a di1 ec tl y prop ortio d
in the time and e f fort o f the s urve yor. l t . s houl
c rease
d eg;;..
the n be the resp onsi bi li ty o f t he surv e y or t o obta i n a
ified •
ree of• prec ision . whic h cou ld b e as hi gh a s can be just
a cco rd i ng to the purp ose o f the s u rve y .

4- 6 . THEORY OF PROBABILITY.
some thi·n g
Pro babi l ity. is defj _ned as the numb e r of t imes ·
r ~e n~e• ~
wil l prob abl y occu r rver the r a n ge of poss ib l e occu
as t hro.,wr
I t is very much invo lved in game s of chan ce , s uch
toss i ng a coin , o r in vario us g&,me s . u s ing card s .
.i ng di ce,
r e~pr ov~ n
T~in gs d o happ e n rando mly or by chan ce · an d thas e a
by pr i n c.ip l es o f math emat ics c ommo ~ l y r efe r re d t o
as prQb a .-
' bi lit y . Di ffere n t theo ries o f pr o b~b i l ity a r e n ot only ap--
.
plic able t o game s o f chan c e , the y a r e a l s o u sed
- .
in . s ei e n~ '
~ r
tlf i -c and eng ineer in'g meas u ~ nts such a s. i~ s urve ying .
Ac c i _d e ntal e rroi s · t= xis . n all surveying measurement s
and : t he i r ipa~n i_tude a nd. fr e que.n cy are g overned. b y the salt\e
g ene r a l princ 1pl e$.A> f . pr~bab i l i ty . I n dea l ing wi t h probab1 ,·
· lity , · it. i s as sumed t hat ~ e re f er pr i nc ipall y onl y ·t o ac:c1,
~ ent~l ' . e rro rs a pd t ha t a ll sy~te ma t i c _e ~ rors and mis~akers
.ha ve · b ee n _ elimi na ted .
effort.
is .
6\ade to el imi na '
I n hi gh pre c 1.s 1on · surveys
t e · systemat
-
ic error s .
s ion of, a measured quant i,ty i s· d epende nt upon the acc i dent
extxa
The p reci-.
-,
a l er r o~ it contai n~ .
• The t he o ry of ·pr o babili t y ts use~ul ' in iRd ica t i ng · the
1>,.reci ~ ion Qf re s u lts · only i n so far as t he ~ _a r e af f e cted by
,_ ac;:cid e nta l et'ro r s . It does n o t , howeve r 1 1n any wa y de t er -
mine the ma gnitude of s ystematic errors ·. whi ch may al s o be
~resent . The t h eor y .a~sumes a ri infinite numbe r of occurren-
c e s of all pos-~ ible eve n ts ; how~ver, it may be a pplied with
9f p.d res ul t s to a 1 imi ted but. fa _i~ ly large number o f obs e r- :
v at ions . To form . a judgment of a probable value o r t he
proba b le pr~~ision of a measured . quantity, it is necessar y
to re l y u pon .this theory . 0
·
. ·· The theory of. proba bility is based upon the f ollowi ng
ass_umptions relative to the ,occurrence s o,f errors :

1 . Sma],l- e:r::'rors- occur more often than large ones and


tha t they are more probable.
2 ~ Large errors happen .infrequent ly and are therefor e
~ess probable; for normally distributed errors, unusual ly
large ones may .be mistakes· rather than accidental errors .
.' 3'. P·ositive and negative e:rro:rs of the same size happen
· with equai frequency; that is, they are equally probable.
· 4 . The · mean · )fan irffinite number of observation s is
the most probable value.

By applying the principles •of probability , measurement s


ro n taihing accidental errors could be adjusted. The most
· pr•obable . · v~l ue of a set of obs er vat ions could then be de-
_tetmined and inherent discrepanci es a:re eliminat~d fro~
_such measurement :s. It must be understood, however, that the
' re~ults of such ' adjustments are not the true values , but
a :i:: e · the mos-t probable values · which could be derived f r om
the given measuremer. ts.
I n this lesson, · only the s i mpler application s of the
l a¥s of t>r,obability will be considered. A thorough under -
standi ng . of it may be obtained by the study of the method
. o f lea~t squares .

4-7. -HOST PROBABLE VALUE .



From t he theory of probability a basic assumption i s tha t '
the mo st •probable val-ue (mpv) of a group o i :re p e a tt!d j,~ cr '
s u rement's -made under similar conditions is the ar i th metic .
mea n or t h e average. Most probable value refers to a q ui · - ;
t i ty which, based on ·available data, has ;-_)_ore chances \
109 cor~ect than has any other
t>e s1nce . the true values of me~~ · . . . · ,,\•)
forever unknown
n,ai~ only approximations
inen s
it . -

uanti ty is measur . o
1:
. ured q~ant1. ties must re ,.,.
iossible to obtain by· measure - - ·
rue values .. • There~ore, :'. i -f a .
given q . ed more than one time ;resulting in·.
more tha n one value , not _.a ll the.· d.erived values are c ~r -:-
,ect • Ther~ 1 ~ only on~ . correct value that shou.ld be c ons·i ..:
aered a nd 1 •t is determrned by · usi'ng the following equatiQn

' mp V =. X =
.
~x In = (X
I
t X + X +r
2 a ;. +X)/n
n

\iher~ .~ v 6r X. is _ the most probable value of ·t l')e, quantity


measured, ~ X 18 the sum of the individual measurements,
and n is the total . number of observations made . . The above
eguati?n is . derived from t h e princ.iple of least square~.,.
~hich 1.s based · on ._th·e the o ~y of probabi lJ ty. · · • ·
In the case o f re l ated measllre men t s tak e n under:: identl -:
cal condi t ions wh ere the . sum . should equal·
exact quantity,
a
ma thematically ·
th'e mos t p·robable v a lues· ar e the observed ·.
values cQrrecte d by a n equ a l ~ar t o f th~ t 6tal ~rror. · S uch
a. s ituat~on wo ul d o nl y -be poss ib le in the ca~ e of angJe~
about a po i nt _o_r angles in a g eome tr i c fi g ure·· . . The · c·or:rect-.
io'n is no t _in p ro port ion •to the ma g nit u de of th e individual
Jneasur e men ts but _ t o the . number of · rela t ed me a sure me nts • · ! \t
is a lso a di f ferei:it ca_s e · for .re late d meas u;.eme nt~ , ·taken
under ide n t:'ica l cond it i ans when the s ·u m should equal to a
s ingl ~ :me asurement . H~re, · the mqst pr o babl ~ !~al u ~s a r~ ob~
ta ined by dividing · the discrepancy ti~e ., ~ the dif f e r ence
betwee n two measured values . of- the same quanti ty) : e qua l~y
~mon~ a l~ the measurements, including the ~um. If_the c o~; .
re ct1on is subtracted from each of the rela te d.' measqr e-
me nt-s , it is added to the measurement repres ent irig tbe.ir
. .
sum,,. a nd vice versa. . .
·
. Determination 9£ the most probable value from a se:;z.:ies.
of measurements is -the principal use of the the ory of prob-
abili ty . . .

. 4-8.: I L L (.!STRATIVE PROBLEMS .


I. MOST PROBABLE VALUE. A . surv~yino i nst r u~tor ~ent -. out six ..
gtoups o f st udents to measu·re a
distance between t wo , points · mork·ed on the
oround .The stu dents came up with th·e fo11 o.w l ng s ix di f f er ent ·v9lue_s: 250 ··
,:25,. 250.15, 249 . 90, 2? 1.04, 250 .50 , and _2 5 1. 22 mete rs. A ss_uming. these .
V0IUitS are equa II y .re li able a nd th a t var la t ions r e~u It from ~cctde{ltOI er-
rors, . d~termln
~
e th e most . pr ob ab le ~ v<Jlu e of .th e -distance measured.
. . ~

mpv or x x, x·
= :E X / n · = c + X 2 + X 3 +· X 4 -t 5 + Xs >/n . ·. . · · ·. : · .·
= (25 0 .25 +2 50.1 5 + 24 9 . 90 + 251 . 04 t250.50·+251.22}/6 ~'_1
= 250 .5 1 m (th e ff\ost probabl e value of the distci_nce measurectt: ~
' :1 .
' .

2 •• MOST P R OBABL;E VALUE The an(Jles ab.out a point ·Q ' have , ·tt'!e i . ·
I ' ' . values. .,130t I ' H
..f o.~owlng 15 2_0 ,142 37 ' 30 II ,an d 87°07' 4C t .~.;,:-
. >ll ., De8-.
O
observed
mtr\e the most probable · value of -each angJe .

I/IITlfOO~CTI0/11
TO SURVCr/1# -
lllllil
tqTllt
39
., 1·
Hg . ~-z. Ati.gl.u 111~.a.~.ud about a poW.

6/VBn.~
~I ·a ··130•rrs•20"
-€2 • 142•37' 30"
~ = · a1•07' 40"
3
. r, = 3 •<number of obHrVtd , ongltl)

Solution :
a) Oetermln lnQ t he Correctlo,n 'to be Applied.

Su~ : e, + ~ 2 + tt3 : 1,0 ° 15 I 20 11 +. J 42~37 1 30 11 + 87°07 140 11


= 360°00 13.0 11 ' (-sum of ·t:h t an.glee obaerved. about point Q)

Disc = 3 60,0 ' '...; ,3 60° 00 1 30 11 •


,= - 3 0 11 (discrepan cy In ,the observatio n)
. . . .

Not.fl : The $ Um of th e angles ob~erved about po/ r,t Q Is subtraot ed f rom


.
.360° ( the- ,e xpected corr•ct . sum)
to determin e ffJB d lscrepan cy.
Corr · = DI sci I n = - 30 11/ -3
2 · - 10 11 (correcUo n to be subtracted from eacn observe~ an,gli )

b) Dete rm inlng the Mo.st . Probat.ie Valu·es .


e-\ . e- + .Corr= 130°t5 20 -
1
1 11
tf"
= I 30~ 15 1 1,0 11 (most probable value of -e-, >
~ •2 - &.2! Corr :s 142°37 1 30 11· -10 11
= I 42° 37 1 20
11
( most pro·bable va l ue of~ -&-2>

=
•i)-
! · corr =
· '9-~
1
3
11
. B 1°01" 40 11
- O
·1
11

= 87° 0:7 30 Crno.• t probable value of ~~)

c) S olu(lon Chfi("k:
-e-'I. ~. e}." + · 4';; ·,= 366°00'
. . 00 11
13 0 ° 15 1 10 11• + 1·42° 37 I 20 11 + 87°07 1 30 11 = 360°00 1 00 11
360°00' 00 11 36 0°00 100" (chec ks')

No t e :- Since t hl two quanti t ies are equal, .fhe above ·solution is . ass um•d;
to bs. correct. ·.·.

~ . MOST PRQBAB I.E 'VALUE. . the observed interior ang l es o-f a tri-
1 11
angle . ore A = 3'-5° 14 37 , B = 96• 30 1 09 11 , and C = 48° 15 1 0 5 11 • Determine
th e d isc,epahc y fat the given observa ti on and t he ·mos t probable value of
e a ch angl e.

L!:~--
40 tll1"1f00/JC1"/0II
rr, S/Hlfl#' .. , •••

Glv•n :
A, B, & C • ve rttc •s of Hl onot•
ABC
n • 3 ( numbe r of obHn,• ct
• ongl••>
/
A<?,"i~•~~•3~1~ 11 _ _ _ _ __ __.- 7c

Solut io n:

a) 0et erm intn9 the Cor r ectl on to be App lied .
Sum 1 • A + B -t C • 35° 14 ' 37 11 + 96°3 0 109 t + 48° I 5 1 05 11
1
1
s 179° 59 5 1" < Nlffl. of the ob u rv ed In ter ior ang lH of trlono l• ABC }

Sum 2 • • ( n ~ 2) I 8 0 ° = ( 3 .. 2) I 80°
- 1 eo·oo 'oo" Hhe correc t cum for th• Inter
• of a t h r•• -
i or an o les
s l d•d figure >
Oi se s ± (Sum 2 - Sum 1 )
• ± < 100° - 179•5 9'51" >
- _+
_ 09" (discre pancy In the observ a ti on )

Corr • Disc /n • + 09 11/ 3


:a + 03" (correc tion to be added to each observ ed angle)

t> ) Deter m ini ng the Most Proba b le Value s.

A' .• A :!: Corr ;: 35°1 4 1 37 11 + 03 11


1
35°14 40 11 { moat probab le value of angle Al

B' • B ± Corr = 96°3 0 1 09 11 + 0 3 11


1
= 96°3 0 12 II (most probab le vol ue of angle Bl

c' • C ! Corr ::: 48°15 I 05 11 + 03 11


. : 1
48°1 5 08
11
C most probab le value of angle C )

c> Solut ion Chick :


A' + 8 1 -+ C' = Sum2 .
• 35• 14' 40 11 + 96° 30' 12" ·1- 48° t-5 ' oe" = 1 eo~ oo ' 00 11
1 11
I 80°0 0 00 = I 80° 00 1 00 11 (ch eck s)
Nott1 : S ines ths two quant it l ss are equa l, t he above solut
i on Is as-
sums d to bs corre ct .

4 . MOS T PRO/ !AB~ E VALUE. -


~eq_s ur emen t of 1hree h~r iz~n tol ongl e5
{see accom panyi n g f i gure } aoou t a potn tf> ore : APB = 12 0
3 1 5P , BPC =
3 7°29' 20n • and C:D = 4 7\3J,1 30° . If the meos ureme n1 of t h8
sing le ,engle
APO i s 97° 37'00 1 , de1er mi n~ t h e most proba ble v a l ues of the
an g les .
Fig. 4- 4. Re.lated ~eJlle.Jt.U.

0
Glv9n: 4 1 11
c:Ct . : A~B - - I 2 • 31 50
1 11
cC2- : B PC : 37°29 20
te3- = CP.D =-47~ 36 30
1 11

cC4 :- APO' = 91°. v 1' oo"


2

· n : 4. ( num be_r of observ ed angles)

Solution:
1

a) , Determining the Correction to be A pplied.


1 11 1 11
S~m 1 = GC 1 + oC2 -t, · GC3 - 12°-3 1:'~0" :+ 37° 29 20 + 47°36 30
1 11
= 9 7° 3 7 4 0 (s um of t he angles · observed a1>o ut p oi nt P)

= -±c ce 4 - S um,> = :t <97° 37'ob'i ~ ,1°37 40 )


1 11
Di·sc
1
= -- 4 0! r ep0ncy in the t_wo sets of obser vat i on
( dlsc1 made) ;
11
Corr o'i sc /n. = - - 40 / 4
= - I 0 11 · ( amount of oorrect l on to be opplled) ·

b) _De termini~g the Most Probable _Vol-ue.

ce , I
= _a: 1
± Corr
11
1
= 12 °3 I 50 II -- IOU r
= 12° :3 J 140 (most probable value of angle APB)
.
,cC2
I
= i:C2 + Corr
1 11
= 37°29 20
1 11
- I0
11

= 37°29 10 (most probable value of angle BPC)

lC~ = cC3 £ ·Corr


1
= 47°36 30
1 11
- IO
11

= 4 7 ° 3 6 20" (.most probable value of angle CPO)


1
' ~
11
= ! · 1Corr11
cC
. 4 = - 97°37 00 t IO"
=· 9 7 ° 3 7 .1 0 (most probable value of angle APO)

Notes: a) T he correct ion"'·s subtracted from each of the three angles,


rxJ1, ~ 2 , l£:J, since their suin e)(,peeds the siT}(/IB -angle cC4 . ·
'-
\ . ·bJ T-o angle ce 4 the -corr_ection is addsd since its value is less -;
than the' otf1er three ·angles.

c} Solution Che.ck:
I 1· I I
cCt _ +• oCz. + GC3 _= cC4
_,2°31 ' 40" -+ 3 1 °2 9 .. 10". + 47 ° 3 6' 2 011 =- 97 ° 3 1 ·' 1-0"
1 11 1 11
• 9 7°.37 10 = 97° 3 7 10 (checkst
'\ .
Note : Since•ths ·tw.tJ qi.Jantlt/65 (J r(J equal I the above . solution is ossum,d,
t t> be corrsct. · • • ·, ·
••
LL:~---
42 IN rffOOU CTIOM .
ro $URV£YIII! .
••.
'
5- 1. RBSIDUAL .
th
The residual, which is sometime s · referred .to as e··d~v1a ... '
tion, · is defined as the·''dif,f er;ence betwe~n any ~aaured
~4lue of a quantity and its most pr~~able val~e ,or .
. . .,..X : . . . . . . ~ -~ . • . . . . EqJ I ) .
V =. ·x·
~
.
I • •
; .· ·. .
Where v is ·the · residual ln any measurelri~nt, X is ,a. me~sure- ·
ment' maqe of ., a particular quanti:ty, .· and X_ is th e most P.t o-
bable value of ·the quantity ~easured • . Re~ .1 d.uals an~ errota.
~r ~ theoretically identical. The only difference 1s th~£
residuals can be calculated whereas errois cannot because
-there iS: . no. way of knowing / true va_Lues .. For a part!-cu,lar
a;·et o·f measurements it is importa nt to compare the resi -
duals with the average· value for those residuals. When v·ery
l a rge ·res.iduals are detected they a r e u_s ua lly ·di'scarded- and
the reg~ired calculations are corit1nued oniy wit~ the re~
ma~ning _ones.

5-2. PROBABLE ERROR .


The probabl e .er r or is a q ua n t i ty which, wh e h added to and
s ubtr acte d fro m t he· most pro bable val ue , def ine s a . r ange
within wh ic h t here is . a 5 0 perce n t c ha ~6 e tha t the - true ·
va lue of the measured gu'antity l i es inside· (or ou tsi d e ) the .
·limi t s t hus set. · · ·
If errors are arrq ng ed in o r der of magni tude, · it will
be · possible to determine t he probable error. Thi s is the
~rr or . that would be found in the middle· place of t h e · ar-
rangement , such that one ha lf . o f the errors are g_rea ter
tha n .it and t he ot h er half are l es s than i t.
, The _ . value of the probable erro r .i s · .calculated wi th the
use o f t ~e followi11g form u·lae wh i c h are der i ved fro m .· the
method of least sq uares.

2
PE= :!: 0 . 6.745 ~ ~v '
·• $
n- 1 · EQ.(2~

PE= ± 0 . 6745 - ~
m . 'J ~ · EQ.(3)

Where: PE, = pr obable error


. . of any 5 _.1·-n gle me a s.ur e me.nt
of a s er ie s

PEm = pr obabl e error of the me an


2
·~~ = summation o f t he s quares of the r es i_guals

n = number of observations I
The dete rmi na t ion and use o f the proba b le err or in s
veyin g i s prim aril y to give an ind ica ti on.'o f t he preci s
o f a parti c ula r measu r emen t. It i s o ft e n a n a pp r oxink
va l ue and i s n ot - ma thema t ic a ll y exa c t si nce only a f ew re-
peate d measu remen ts ar e us ual l y made and th~ co nditi ons of
measux:e me nt a re not rigid ly c ont rolle d. For . e xamp l e , if ·
23S.5 0 ,, m repr es ents the m~a n ot ~ost probi ble val ue . of
s evera l me asur ement s and 0 . 10. m repre sents thE!' proba b l e
er r or of t he mea {l value , the cha nces a re even t ha t · the· true
value l i e s b e t wee n 235.4 0 ,m and · 235 . 60 , m , as .i t i -s a l so
proba b l e that the true value lie s outs ide of these . li mi t i ng
va lues . Not e t ha t the lower and up per l i mit s ar e det er min~d
by corre spond ing l y subt r a cting and aOd ing 0 . 10 m .to 2 3 5;50·
me t ers . To expr e ss th e proba ble l i mit s , o f pr e cis i o n fo r '
th i s parti cul a r c ase , the qu~nt i t y sho u ld be writ te n as

2 35 . 50 ! 0. IO m

The s i g n or direc t i on of the proba ble er ror i s not


kn own a n d there fore no co!'."r ection can be made . It do es ndt
spe cify the magn itude of t he actua l erro r , n or does .i t .in-
di cate the error most likel y to occur . The proba ble e rro r
is not a subje ctive _guess . .I t is a logic al estim ate bas ed
u po n the metho ds and equipm e n t used, upon the exper ie n ce
of
the obser vers, and upon the ·fie l d cond ition s exist i n g du -
ri ng the measu remen t .

5- 3. RELATIVE ( ERROR) PRECI SION.

The total a mou n t of e r rbr rn a gi v en measu reme n t shoul d


r e-
late to the magn i t ude of t h e measu red q u antit y in order
to
i ndica te the accur a c y o f a meas ur emen t. In surve ying mea -
surem ents, · ratio of the e r r o r to t he measu red quan tity
is
used to defin e the deg r~ e of ref i n eme n t obtai ned. '
Relat ive error , somet imes c a l led relat ive preci sion, i s
'expz= essed by a f_ract i on h a v i n g th e magn itude of t he er~ or
in the nume rator anq the magn it ude of a measu red qua n tity
i n the denom inator . It i s ne c essar y to expre ss both • quant
i-
ties in the same u n it s. a n.d the numer ator is reduc ed to
un ity or 1 in or~er t o r~o v i d e an easy comp ariso n wi th
ot h er measu remen ts . Fo r e x ample , ' if f o r a parti cular mea-
sure-: .ent · the proba ble error of the mean is 0 . 10 m and
t he
111os t proba ble value of the measu remen t is 235. 50 rn, - _t h e
relat ive preci sion (RP) would be expre ..ssed 2s 0 . 10 / 235 . 50
or 1/235 5, ~lso writt en a~ 1:235 5 .

5- 4ft WEIGHTED OBSERVATIONS.

It i s·not alway s possi ble to obtai n meas~ reme nt s of equ al


re li abili ty under simil ar · cond it i ons . Man y s urve ying mea -
su r eme n ts are •made under diffe rent ci r cu msta nc es and condi
-
. .
t io ns a n d t heref ore have diffe rent deg r ees o f · re liab i lity
' ' . .
The prob lem ofte n enc oun tere d is how to com
bine ' lhes e me a ,
sure men ts and dete rmi ne the mos t prob
able valu es. For s uch
a situ atio n i t is nec essa ry to esti mat e -ae
deg ree of re ,
liab ilit y (or wei ght) for each of the mea sure men
they are cc,n bine d and th'e . mos t prob able ts befo r~
v alue s are dete r ..
min ed .
T"he assi gnm ent of rela t ive we i gh t to diff er
en t mea sure ,
men ts is usu ally _ base d u pon the judg men t o ~
the surv eyo r,
the num ber of mea s urem ent s ta ken fo r
a par ticu lar qua ~t i t y,
and by as sum in9 tha t the wei ghts are inve
r~e l y prop o~t i o n-
al to the squ are of the prob able erro
rs . Amo ng exp erie nce 0
surv eyo rs ,. th e assi gnme nt of we igh ts
to obs erve d valu , s i s
ofte n a mat t~ of judg men t . In some inst anc ~s we igh t s
assi gne d a re
on the ba-s is of ~ea ther con diti ons pre
vail ing~ ?t
th~ time the mea sure men t s were made .
A ~ea sure d leng th ·ob-
tain ed on a brig ht earl y mor ning cou ld be con side reu : as
more reli abl e than one mea sure d o n a cold a~d rain y
Thi s m, t hod of assi gnin g weig hts will d ay .
req uire good judg men t
and can oniy be exp ecte d of more exp erie
nce d su rvey ors . In
the case of re pea ted mea sure men ts, if a qua ntit y is
sure d, mea -
for exam ple, i n two rep etit ion s by grou p A and in
four rep etit ion s by grou p ·a, then the mea s u r e men t t ake n by
grou p B shou ld be give n twi c e th e wei gh
t of t h e mea su reme nt
of grou p A . This mean s t hat the me a su r e me n t o f g r
rega rded as t w ice as r el i a b l e as that ou p .Bis
o f gro up A. Th e as-
sig nme nt of wei ght s i g p u rely rela t i ve. For this stat ed
exam ple, any weig h t s in t he rati o of 2
to 1 may be ass i gn ed
in.s tead of 2 and 1 . For ~nst anc e, lhe weig ht s may be - 1 and
l/2~ 4 and 2, or 1 6 and 8 .
• •
S-S . INTERRELAT I ONSH I P OF ERRO"S .

In some ins tanc e s it i s requ ired to dete


rmi n e ho w th e fi n al
resu lt is af fe ct ed ~hen a com puta l 1on invo l v es
qu ant itie s
th~ t ~re ?ub ject to ac c1 dPn tal erro rs.
T wo ~ o mmon ly app l i ed
pr1n c1p les of the th eor y of erro rs invo
l v e t h e s umm ation of
erro rs and t he.p r odu c t of ~r,o rs . Thes e pr in ci ple s are gi -
ven to prov id~ the s tud ent a uett er und
erst a ndin g of t he
prop aga tion of erro r s.


1. Sua aati on of Rrro z s_ . I f s e vP rul mea
s u red q uan tit i es
are add ed, ea c h o f wh ic h i$ <.1t teclc d b y acc lden t a
the l err o r s ,
prob able '. e r ror: o f th e s u m t~ qi v e n b y
t he s qua r e roo t
of the sum of the sq'.'a re s ot UH:: ~e p<.1
ra tt! prob ab l e err or s
aris ing from the s e v era l sou rces or

PE = + .. I .·lE 2 + PE 2 + 2
PF-3
21
I ~~ I
1
2 + PE Eq.(4 )
n
..
Whe re: •
prob able erro r of the sum
• •

PE11 f'E2, etc =• prob able erro r: of each mea sure men t

, 46
~ -~-- --- ,,,r,.oou
~ c r,011
wttvcr,1 1•
n = n u~er of val ues added

2. Pr oduct o f Errors. For a me asure d q uan tity which is


de t ermined as the prod uct o f two othe r ind e pe nde n t ly mea-
s ured q ua n tities suc h as, Q 1 and o · ( wit h t h ei r c o ~ r e s p on -
2
din~ pr obable e rrors) , the p r oba ble err or o f t he pr oduct is
g i v en by the fol~owi n g equati o n

PE z ! ~( 0 x PE 2
) t (Q x PE .)2 1 ·· · · · · · · · Eq.(5}
P 1 . 2 2· I

Wh e re: PEP = p r o bab l e e rror of t h e prod uc t

o, 8 02 = measured qua nti t ies

PE 1 a PE = probqble er r or co r respo nd ing to ea ch


2
quantity measured

5- 6 . ILLUSTRATIVE PROBLEMS .

/. PROBABLE ERROR . T he follow ing val ues wen~ determin ed i n _ a


ser ies of tape measu r emen ts of a Ii ne : l000 . 5 8 , 1000. 4 0 , l0 0 0 . 3 8 _, 1000
: 4 8 , 1000.40, ~nd IOOQ .46 me ters. Det erm i ne the fo ll owing :

a) Most probable value .o·f t he mea sured l en gt h


b) Probable e rr or of a s ingl e mea suremen t and pro b a b le er ro r of the mea n
c) Fina l exp r ess ion fo r th e most p ro b able length
d) Relati ve p r ecis ion of .the measu re me nt

Solution :
a) n = 6 (numbe r of ob:ie r v oflo ns)
• -~ X = X1 + X2 + X3 + X4 + X5 + Xs .
= 1000.58 t 100 0 .40 + 100 0 .38 + 1000 .4 8 t 1000 .40 t 1000 . 46
2 6002.70 m

mpv or X= :EX/11
= 6002 . 70 / 6
= 1000. 45 ( m ost pr obobl& value of t h_o mea s ured le n g t h)


b) v, = (X I- X) = 10 00 .:5 8 - 1000 . 45 = + 0. 13
v2 = .( X2 - X) = I 0 00 .40 - 10 0 0 . L't5 :J ·- 0 . 0 5
V3 = (X 3 - Xl = 1000 , 38 - l000 . 45 ::, - 0 . 0 7
. V4 = ( X4 - x). = 1000 ,48 - 1000 . 45 :J t 0 .03
V5 = ( X5 - X) = 1000. 40 - 1000. 45 = - 0 . 05
V5 ::; ( Xs - X) ::r 1000. 46 - 1000. 45 = t 0 . 01
:i. V = 0 . 00

..vl - ( - o .C5 )2)


\12 -
~
=
("+ 0 . I 3 = 0 ,0 I 6 9
2 = 0.0025
v42 · = ( t O . 0 3 )2
vg
2
0 .0 0 0 9
= ( - 0 . 0 5 )2 =- 0 . 0 0 2 5
v2
3
2 .
= (- 0.07 ) = 0. 0049 v; = · ( + 0. 0 I ) 2
= 0 . 00 0 I
'llli:!
~
v2 = v,2. + . v22 + v32 . +. v42 .-t . v~ · + V:2
5 - 6 . . , .
= o.0169 t 0 .0025 -t 0.004 9. + 0 ~0009 + 0.0025 t 0.000 1 ·
= 0 .0278
Tab'uldte d S o/ufir,n.:
t 1
MEASURE D
LENGTH
CX) .
RESl bUAL
( v.: x-x ) . - . SQ OF R\S IOUAL
(V ) ,
: .. ,
1000 .ss m +o . ·13 . 0. 0 1.6 .9
. I 060.4 0 -O. Of5 0; 0025
I 000. 3 8 - 0 . 07 o. 0049
l000. 4 8 t 0 . 03 0 . 00 09 '
ropo.4o -o. 05 0 . 002 5
1000. 46 +.o. ot · 0 . 0 0 01
2
~= 6002~7 0 ~ v= 0 . 0 0 · :e v = o . 02 7 8
. '
. .
PE 6 = ±0.674 5 i ~
'\ ~
+o 6745
- .
~ (6
o .027s'
- l)
= ±0.05 m (p ro bable er ror of a sing le measu re men t)

1
2
PE = :t 0.6745 _{ ~v . = ± o . 674 5 "'l o.o27 ~ ·
m ~ .n ( n-l) . ~6 ( 6-i)
= ± 0. 02 m. , (p robable er ro r of the p, ean)

c) Th e ref ore the Ieng th oJ the measured Ii ne m ay be ex pr essed as IOOO


. 4 5 m ± 0 .02 m. Th is m ean s that th ere is a 50 p erce nt chance th at t he tru1
d i '\tan ce measured pr oba bly fa l ls bet.we en 10 00 . 43m and I000 :47 m , a nd
th at it s most prob able v a lu e is l000.. 45.m. Ther e is also ,,howeve r; a 5 0
per cent ch ance tha t· the· t r ue d i sta nce l i es ou tsid e t hi s r a nge .
.
d) RPS = PE 5 /mpv · :: 0. 05/1000 . 45
- I/ 20,000 (th e r e la t Ive pre c l s Io n of a single measurement)

RPm = P~/mpv . = 0 .02 I I 000 . 45


= I / 0 1 000 (the relative precision
. f
of the mean)

2 . 'WEIGHTED MEASUREMENTS. Four measurements of a dis,t an·ce were


re corded as 284 . 18, 284.19, 284.22, and 284 . 2.,0 meters artd gi ven weights
of I, 3, 2 , and 4, _respectively. Determine the w~ighte.d mean:

MEASURED A S SI GN E D '
LEN,GTH WE IGH T p = X CW)
( X) CW)
284 . 18m I 284 :1 8
28 4 . 19 '3 85.2.57
284 .2 2 .. 2 568 .44
~84 . 20 4 11 36.80

Sums
.. . · :i.W :: 10 ~ P:2841 .99

~p 2841.99 ( mo st pr ob able valu e ~


We ighted Mea n = 284.20 m
~w 10 the dis t ance measurtd

. 48 1NrRoo ucr10N
L .. ~ - - ro SUR VEY/N(J
If Is de sfre d to dete rmln • the mos t pw -
J. WEI SHT ED MEA SUREMENTS. d at d1ffe renf ti mes by d1f -
bobft value of on onote wh ich hos been mea sure . 74•
, ur•nt observers with equ 1. . care . The va lue s obse rved were os foll o ws 11
tour moo sur om ents l. and 74• 39'
39•4 ~"< ln two mto $ur emen t s), 74•3 9' 2 7 Ctn
5 ' (In six mea sure me nts ).
1
3
- - -- .
MEA SURED
- --
VA LUES
NO . OF
08SE R VA TIONS ~
- -~
PA OOU CT OF
-£ TWO _OU":_N_! IT l ~ S
74•3 9 • 4 ~"
- -- • f 4 9 • I 9' !O ,,
2
7 4•39 ' 27 " 4 299• 3 7 ' 49 "
11
7 4• 39• 3~ •• 6 447• 01' 30
- 11
S ums 12 89, •~.q I 48

995 •54' 49"


We i ghte d Mean =
12
1 11
= 7 4°39 3 4 ( most proba b l e va l u e ot rh e .Jng l • meas ur ed)

l s to es tabli sh t he e Jevo t lon


4. WEI GHT ED MEA SUR EME NTS. Lin es of leve e obse rved e le vatio ns of t he
of d po in t are ru n ove r four diffe rent rou t es . Th
rmin e The mos t prob ab le vol u
point w i th prob able err ors are ~ive n below . Dete
of the elev ation of the po i nt .

RELA TIVE
I
OBSE RVED PROBABLE I P: EL fV ( RW )
LINE E:...EVAT ION ERRO R ( EJ2 W2- WEIG HT
CE ) 2 (RW}
( EL EV) CE> '
}
--f- 16 . 00 3517 . 3 12
I 2 I9 .832 m, ! 0 . 00 6 m 0 . 0000 36 2777 8
4 . 00 8 7 9 . 72 0
2 21 9 . 9 3 0 ± 0 . 012 0 . 000 144 6944
39 1. 068
i 0 . 018 0 . 0003 24 3086 I. 7 8
3 2 19 . 70 I
2io . 02 1
4 220 . 0 2 1
-- I ± 0 . 024 0. 0005 76 1736
Suma
1. 00
22 . 78 5008 . 12 1

igne d lo line 4 sin c e i t hes


Note : A rela tive we iq t,r equ al to 1.00 i s ass
f/1 8 sma l l est we i 9ht f W) , a n d t he r elat ive
weig hts of lin es I , 2, a n d 3
are comp ut e d a s f ol l ows. · ·

RW 1 = w11 w 4 = 277 7 8 /17 36 = 16 .00

RW2 = W2 / W4 = 6 944 / 17 36 r 4 .0 0

3 086 / 17 36 ::;
I 78
RW 3 = W3 /W4 =
~ ( P) 5008 . 12 1
We iohte d Mean r. ~ (RW) = 2 2 . 78
e te v o t 1oo of t he
:a 219 . 847 m (th• mos t proba ble valu e- of t h e
po int)

6 . WEI GHT ED MEA SUR EME NTS. The le noth of o l i ne was mea sure d r e-
t he prob abl e err or of £ a rh m ean
pea ted fy o n three diffe rent occ o si ons and
t s:
va lue was comp uted w i th the fol l ow i ng resu l
.
I.st Se t of Mea s ur eme nt s = 12 01. 50± 0 . 02m
2nd Se r of ~1eo sur em en fs = 12 0 I . ~5 ! 0 . 0 4 m
3rd Set of Mep s ur em ent s 120 l. 62! 0. 05m

U t te rmin i3 t he we i ghte d mea n o f the three sets of m r o s ur e m enr s


. RELATIVE
........
MEASURE[) PROBABLE I
MEASU REMENT VA L UE ERROR ( E >2 W•~ WEIGHT PsXlftV,)
(X) (E) (EJ CRW)
6 . 2~
--....
a 1201 . 50m !0 . 02 0 . 0004 · 2500 1&09_ 3 ,
b t 201. 46 t 0 .04 0 . 0018 62 5 l.~6 1874.21
e 120~ . 62 ! 0 . 05 0 .002~ 400 1.00 1201. 82
sum• 8 . 81 --.... ,
1oae5.ie
Not e: A relative weight eqtla/ to I.. 00 is assigne_d to me asurement c sinc8
i t has the smallest weight (W), and ;he _relative we ights of a and b are c°'"4
puted as follows :

RWO = Wa!Vvc = · 2500/400 = 6 . 25


RWb = Wb / Wc = 625 / 40tJ = 1. 56

We i ghted Mean = ~ (P ) /(RW )


= 120 1. 5 1 m
= 10585.26/ 8 . 81 .
(mos t probable v a l ue of t.he li ne me o a u re d)

6 . SUMMA1'/0N OF ERRORS . The th ree sides of t r i_a ngu l a_r-shap ed trae1 a


of lon d is g iv en by the fo l lowing meas urements and c orr esp ond in g proba bl e
err ors : a = 162.54 ± 0 .03 m , b = 234 . 26 ± 0.05m, atld c = 19 5 .70 ±0.04 m.
De t ermi ne t he probable error of the sum a nd the mo st pro b able val ue of
th e p e ri me t er.

PER= a i-b + c :i 162 . 54 + 234.26 -t- 195 .70


:i 5 9 2 .5 0 m ' (per imeter of the tr o ct of I a n d)

PE = ± ~ (PE1)2 + (PE2)2 + (PE3}2 '


' 21
2 2
= ± ~ (0.03) t- (0 .05) + £0. 0~)
= ! 0 ~07 m ( probab l e e rr or f or t hfl sum of the t ,hr ee measurements)

Note: Therefore , the per i me t er wo u l d be expre ssed as 592.50 ± 0 .07m. Th/J :


means t hat the true length of the perim eter p r obably falls be f wefin 592.4:Jn. !
and 592 .5Tm. :
I

7. PRf'DUCT o/: ERRORS . The tw o s i des of a rect an g ular l ot were meo·


su re d with certa"1 estimated p r obable erro r s as fol low s: W = 2 53. 3 6!0.06m
an d L = 6 2 4 . 15±0.0Sm . DetePmine the area of the lo t an d t h e probable
error in t he res u lt i ng calcula ti on . .

Area = L x W = 624 . 15 (253 . 36 )


= 158134 . 64 sq m ( rrea of the rectangul ar l ot)
1
PE p :i ! ~ (L X 2
P Ew ) + ( W X PE/
2 21
= ! ~ (624 . 1'5 X 0 .06) + (253 . 36 X 0 .08 )
= + 42. 58 sq m (p r ob abJ & er r or o f th e calc u l a ted a r ea )

"f_ot'e: Therefore, t_he area of the lot wouti be expr e ss ed as / 58 J34 . 6 4


...;. 42 . 5 8 sq m . Thi s mean-s that th e t ru e ar ea of the /o f p ro bab l y f a ll s bef-
ween I 58 C' 9 2 . 0 6 sq m and I 5 8 J .~ 7. 2 2 ..sq m.

Вам также может понравиться