A MATHMW TICAL MODL OF GIEOPHYSICAL EXPLORATION FOR ELONGATED ORE DPOXSITS by ES ROGER C.GORE Geophysal nSieer Ceorado sool of Mine SUBMITTED IN PARTIAL ULFIU LM OF THE RtTWQUI(r.EIMTS FOR DEGREE OF XASTER OF mE Solnfor at th MASSACHUSETTS INSTITUTE OF T EIHIOLOO 5tedbo?1 Signature of Author. Departent oft Certified by............. Aocept4 by............. .p 1958 .. v. r.,..r.. 4 sand Geophystes, Ausget 25, 1958 04 ,-.WV .. W., r C1Irman, artuetal # .... .. dtee an Graduate Stgnsxts • AB3TRACT A MATHEMATICAL MODEL OF GF ICA$ EXPLORATICS FOR ELOGATED ORE DEPOSIS by Roger C. Gore "ubmaitted t the Dear A aot 2S, 198 fol the te~e the itpattil ofMaster itmnt otf ol fitlMuent f Saeenes, In an attempt to establish the batsi lte of pbnleft elnaif t t ai. Geophystes on of the requtiresents eeneeptv for lAn erived wh-h P general MatheMatlal model 1to eMaeted return fromaa ertan &MA of ez a et the *aptta spent This goeral fl-el is a ltd to the problem of sarbing for elongated oebtdint with sostant stke Q n s rgiton in whi h Virtless deposits give tti l gnphyseal realtshe ee osed is sea to provide a perfe*t eat a ot the gradle/ ength etf the rebodies ant it to asmsed that parallel saro h lAeN pepeaular to the strike are used. x in **ats are seamed to be tinar ft etiet of value and ill* tas ects are aasued eenstant for p t!VitrtIezlr type of depeost. of the umrre shoving the expeoted return for various values ime spacing and 1mnima aecepted gme/ /ength are oeanted for the ease of an exploratioa Pgram New Br~drwiek Canda. The ret its ttea~e I it n noisy areas the optiai line spsaing is equal to the maxiamu posible iagth n of orebodies in New Brunaswik in less tqy reas any line spaoing will be poftta0 eand an the average but the eptima line spactgs al Z)as close tgether As posetsbe, I abtn the a a psible length of orebodtes. ThDesis 8 pe okiert . Dr. Patrick P4. uthrlq Titlet Professor of Geology TABLE Or OQT0 Mtj~mttili Ft' datta of fle RottrI flww0tom Co lot.A' flpr0*elon Or a.o R.tnn= POIn AMfltntMn Of the Oo~pte MAo4 to Mtb0*n ** **.**#* G~b7SI1 ?WRS*t0 II NW '9.. **2 *estq*#*o..0**#.e*0.4t#***3 Afl~2t1 *e***4* Z *e*o~a~ge.*q*o*#*#qoeqeeqj Ol WI A itgl? a se bes w1e"OeO ?te, t wee" "t Aupete a grat 4Aat bt of e 1 As new i he oas nOa, be!as dae th Uatoae has beato co'ci treosent aend eaetp i to e ewapIuS nowmy Ua of *roble haO 4a4:tay bee d U Uapet i ent Lphys new ert of to the bti taes8t 840 t Wt theats thfnt to be on ep*Untory efftrt p es n eiUtd to 4eIapt$ nreeffiat e4 a be ooeitere theasutaen ofa e ewloratio etwevetr, i orer to kehp tthis thesis from evel-O tag to a hi y abfte lt rehaa of om seass, the enerat probs of eaptlratioan eoffinp 4 to tho proble with fted strite. of the dtitoultt problem. but the of seareohing for -3.1 has bee spetial ated. deposits this spetiw problem includes say easeount oerd In solving nr* attrtst 446Ama ties invelved to ecasiderably Ins intriate: a destrable qauaity in a preliminary work. Every attempt is madoe in this thesis to solve the parti. otaur proposed probeal this thesis has nt the sin hope in writing ,oweer, bee to develop spaifle solutions to apete problems, but rather to deelop a broad en*qp l base ftr the solution of s* similar problems en**toed in the pnag of an exploration progra. Far te reason, htirever neosesary data vas not available, 2 the necessary ftcts were hypothesized as closely as possible from nast experience in order thrt a total modtel miiht be coxoleted to t t ost if the general a pVroa.h till produce a result conformnble with expcrience. WIERAL MODFL AW attept to enstret or 4devt Ite at w ot prems req ir a. rmstia sir a * *. the peta, ayp ep l S es a s ab a psste s.4ma th n stt$nt.te w eetgUt aese . a oe *taq f 0,4Gwil b* as t a or wlass at the a prtofl a by th. psn tAt th*: mAsl to basst. $eat00 I . suhh a h neG l " s aet kae n aea rp bo-st.arr btSy, tmatrated .xtte as t4trasrtatM athMastty f *flw t Rothe"sanrolistedaaEi avt4 bt Atah a pstoaiarso Ms ail tehe ehsese e be eplord bay a ole oftboar toun PApeitflta ate ti a, ouras ta*f abl iso** 'non area It Airaleo. at Nurther, ve asswn tht the gle hisaw eans r 1m101otor in words. ns.m that 41 tSypes e sati *f* dpnite are equally UZatooL to*ear evr-here In the obese area. V also postulate that another kIA of mterial ocars similarly to the valuable material ea gires a geophystleat respoase inseparable from the response of the valuable deposits but this other material is worthles; to simplit the terminology, deposits contaling valuable material will be called ore deposits and those deposits which contain no valuable material aatal will be refered to as noise deposits. Geophysilt Rypothesee We shall assume that some sort of geophysieal Instruaent whiah responds to ore depests a deposits In a similar ma r is used whicha noise n provide an accurate estimate of the Srade or apparent grade ( if the deposit is a noise deposit ) of the anomalams deposit. Thi instrument is to be transported along equally spaced parallel lines aoross the chosen area perpendioular to the known strike, aed from the response of the instrument certain small, as compared to the chosen area, areas will be selected for further evalurition. EVution Hypotheses We ase me that all anomalies a oepted are evtluated by diamond drilling. The eost of a sufficiently exact evaluation of an anomaly is assumed to be a canstant for all ore depsit anomalies and noise deposit anomalies, howveer, the two constants any be different. By " sufficiently exact " is meant that enough drilling is done so that it can be reliably determined whether an anomnly is assoiated with a ore dposit or a noise deposit and, if there is an ore deposit, whether this ore deposit can be produoed at a profit. C A very itmprtant assuption that we make is that sUftlient funds are available, perhaps from some " benevolent banker , such that =l accepted anomalies may be completely evaluated; however, these funds mIst be repaid and so they are Oonsidered to be part of the exploration eost sa. Productioln lypothesea In produoing an ere deposit we asam that first an initial investment, I, must be made and then for every dollar which is ultimately earned frm the sale of the prodeed valuable material a eertain amout mast be paid for production and extraction costs say , * Therefore the total aost of developing, producing, sad extracting an orebo4y of value v a c.v" the minima 4 We *an then express value of an rebody which it sto profitable to produce from ;,f O , or .,, , or for brevity letting -c, =-a , r,., " In this model the only sourse of revenue eonsidered is that revenue from the sale of the produced material. No attempt s made to include revenue from sale of elaims, stock, or information. QUANTITATIVE WPR SSION OF TH OTRAL MODEL G The general model, as detseribed in the prellminary hypothese, oft an exploration program can be characterized by a mathemateal model whih has three states each state having a partioular assemblage of probabilit- functions. The first state is what might be called the ' natural " state in which the probability functions are these which naturally occur in the chosen area of exploratlon. The three most important probability funtions of this state are: f') , the probability density funation of the value of any dhosen ore deposit; PCX) , the probability ftnction of the total number of ore deposits which occur in the chosen area P(~a) , the probability function of the total rmber of noise derosits in the hoden area, In proceeding from the first or " natural " state to the second or " evaluation " state we must go through the geophysical stage of exploratlon. Ator the gee.' physteal survey certain deposits have been aooepted for evaluation and others have been rejejected; we hope that the more valuable dpposits have been aooepted and the submarginal and noise deposits rejected. We therefore have three different rbp;4ity functions: gv) , the density function of the accepted ore deposits; PU() , the probability funation of the total number of ore deposits which have bee aceepted; yP.,) , the probability funtion of the total number of noise deposits which have been aecepted The transition stage from the second state to the third or " production " state is purpose of the drilling stage is the drilling stage. iThe the rejection of all submarginal and noise deposits since the geophysical stage will probably do an imperfect job of seleetion. The V7- " produotin " state will therefore be characterized by two probability functions: hf) , the density function of the value of all ore deposits which wIll be brouaght into prod tion; and P() , the probability funetion of the total nmber of ore deposits broa t tinto production. Due to to the effect of drilling P(sae for no , in other words all noise deposits hrve been rejected, The question of a criteria of succes of suach a complex operation is of course one open to a considerable amount of variation in solutn soluti and it s obtful if any one criteria could be developed which would satisfy all exploration concepts, To me, the most natural criterion seems to be the total ultimate returnr where I define i, returan as i value * costs. Since we have assumed that all revenue comes from profaeed wealth, the ultimate rlue of all the deposits discover-ed by exploration is ,S '; , the sum of the individual values of the discovered deposite., However we have three oe*t funtions: geophysles costs, drilling costs, and production costs. The geophysies costs I shall represent by C ; the drilling costs are N,Od t 404 , where D/ and P are the cost of a. complete evaluation by drilling an ore deposit and a noise deposit, respeattely; the production costs are, as already d&sOtsed, (c0,") , The Th total ultimate return is thust R,-;- - ;cri - N, -a.. or rM Rra - -n . . , i.-c- I, Rowever, at the beginning of an exploration survey we are concerned with the future return and so we have to consider a probability function for R . Therefore in planning an exploration survey the quantity whieh we will be trying to maximize is the expected retum,, (R) . From the theorem which states that the expeated value of a e expected values ( Wadsworth, sum is equal to the sum of - D eur,) sj -E(N f. 1956 ) , we have: t R) oEl . Unfortuxntely it is imposaeble to let - 0, E(A,)- C. is since N -r £(m 4 ) - -E )H , itself a random variable. There also is no probability function for C since C is one of the parameters whose choice should be indicated by this model. Howvevr, it can be shown that if N Anad v are stochastically independent ( See Appendix 1 ). ) ) E( then EC, 3P9 D E(NM) - Do E C.) - C. ) -'AThereftre; E(It) GI)[a Etrl vE XATREMTI0M FORtAlO N OP THF IRVURN FUpOTNC As outlined in the previous secttion, a complete exploration progrn Ean be divided into thre states: natura, ealuation, and uprodetion. With each one of these states we en assooiate a roup of probabilites which deeribe the 4dstribution of the total madber of ore deposits, the total aber of noise aposits, and the ~Uae of an erbody. ( See fig ) If the exploration program is eonsidered in an abstrat sense as a systea AiEh passes throuh these varias states, then the transition st4ags, geophysi ant Ar ling, sne be eemf$Aered as treastorming rahanisms wbh trans form the probability d1etributions of one State into the probablty distfibutions of the next state. There- fare if the " natursl " state probabiities are known or tasS an ftthe transforming mehanis oan be derivtd or approxtmt," it will be possible to Aerive the probability funetions of the esucceeding states. After the neeessary probability funotions are derived, the return f3ttio E() man then be ftonlated in terms of these probability f~antions. Speoifically frEa (nR)- E( N,) aE r) -A _ - 06, E() - , E(.,) - C () we realize that we need to derive: P), the probability that a ore deposits will sacur in the produation state; P(n) , the probability that an ore deposit which oc rs in the production state will have a kalue v; P(N,) the probability that there will be N ore deposits in the evaluation state; and P(k,), the probability that there will be nnise deposits in the evaluation stte., A Natural State f 6.) , P .) , PC*) Geophysics Stage Evaluation State Drilling Stage Production State M(v), Pi N), P OG.) PHASES OF PT EXPLORATION PROGRAM fig. 1 Natural State Probabilities Since this thesis i soncerned with discovering elongated, needle-haped, orebodies and sine a basie assatimt Is that geophystles an provde a measure of the mrdheof the deposit, It wifl be more eonvenient to consider the probabilities of the length , at the naue per unit I U, A, at then eombine them thaurgh the equation v: A/ to obtain the proability distribution of s wh aeoeslary, than to as sue a probability distribution of Initiafly. The probles n aM A the ~ arises as to what distributions to use for solution of this problem 1t extremely 4UiOUIt tue to the s tty at data. of Minig Seeaaue of the natUw4 retioence Oapanies to release info mtten oonoerning any eamine deposits eept those I prttion, most available infobration oncernet only profitable orebodies art even this infteration is so e restricted in nature. Therefore aw deeision made entirely on avaiable data vwl zwaeessarfly have to ignore the proportaiohlly neth higher nubbr of submarginal and nearly worthless deposits. As this model is an attempt to look at the entire asspeet of exploration for erebedtes, an attempt m st be made to inlude ndat adjust for the*e low value deposits. Sinee data is laeking, the next best approach is to asease a distribution which coneords as elosely as possible with the faets which are ayailable and with what we know to be likely from past e erience. i One tnotion which seems to agree quite closely with what we ight expect the actual distribution to be in nature -9 o S * o. -< oexAX is however the total area under the ourve to finite ant equal to ~ I" t a ensity fntion is then tfonr as , we n have a probability density fctton itich is a mdified fore f the Beta Distribution. The eumlativeletribution F(s) : L'* ;n ? n tion JA s has a verieal elope at the ori4n for c 0 and ftr deoreaszg di has an inereastas probability in the smaller a.es 00*X of ( See fEtotia i . 2) , The , - I < ortes &0 toble ideajy suited for the escript n of the f oeornto of -eettain propertes in nature sUce by propet etioe oft v e *an mfke as lag oar 1as in a proprtion of the total astritaton small a range near zee atIll have a tntiizst the varPiate as we AI* ft probability ftistribtin er entire r'wge of variation. Ntth I MaA seem to tall int the lass of natural variablee whtia might be described by ths Of the eta Distribation and we we shafl let Pg)r A faj *Olt L foAs A ,-I<ers ,- r A0 tfo L (S) iooording to one of the basic hypotheses of this model the entlre area of interest is geoleogicaly honogeneousj this is the same as saying that any given part of the Chosen area is equally likely to contain a deposit, If the further stipulation it mde that the Oocurrenee of one deposit has no effect on the probability of oeQarrenoe of any other deposit, then itSan be said that deposits ocuer individually =nd collectivly at random in the area of interest. If the mean n fber of deposits per unit area Is some constant , then it #an be rigorously shown ( Fry,1928; Operations Research Notes, 1953 ) that the total anuber of deposits ,T t T I, 1.0 "'.' If:!: . . .. ... 1? ..... .. ..... .... F:: 14L- m_ r:L Z ,_ f, . L : : ..... ..... . ... per wait area s thed Potsson Distribut on, i.e., A0or dtiny wo an l thes probability istributton of rtorke deposits th. total PCi vc e as 6,2 '-wo,2i aje athe astural state. toe &adDrtling Probabilities Geop lnphytes an di at theb from e abitties r t t in test te th na bot drilg tna W tie one*Atdia state tate. they bot alter the distribtion of by -sps o agt tszAe Plstrs ensthe pre'tw *tteto ~et r4.eS thapr aw t- to ii the sueeflng state, U *b*eet Mas a apobsilitty Kf at the oerreose s t property the obojet givn the probability of aceptig * at x sto lI)whoMe A referss to the -evwt of p ety te SAoptsaoe, tht p (Ala) pr) z P A,) the probability Pfr etar at aeopting an objest with the property x. or tears of course, P(AIx) represents the *7phys*s drtflin probabilities eat Pt) represents the preceding state probabiliti ess ith regar4 to the gaophysies probabiities, in lines must order to aeept a deOPost one of the tsy first ooss it; the probability of orem"sing a deposit v italnspacI P(Al ej. e10) 1 I A ) For the drilling transition probabilities we have already stipulated that the drilling will reeet all unprofitable deposits, therefore PCA / v, .,) : 0 o C v., () =I v,;, .rL V Noise Probabilities Little information is released oonoernin low value deposits, but even lese informaton is avalable oncerning the ILse, shape, and ==ber of worthless deposits which have been discovered. With so little informtion, asy taomtions we may assin to noise deposit occurrene lie tn the realm of pure conjecture, but, nevOfthless if a complete strumtue is to be onsidered, some attempt mast be made to arrive at realistle fntions. One way to get aroamd this difficlty is to assme that nature oonstruated noise deposits with the same probability finotions as ore deposits but possibly with different Valune of the parameters &eoorinftaly we haves ,O I, P'n.) .- I le f(Aa, J at Cr) - - a. .A,,- < Wes 40 I.SO The geophysies is stipulated as being nable to disoriminate between ore and noise; however, if the maxim length of the noise deposit is the maXate larger than length of the ore d*posits sad if J < 1. then it would be possible to re ject all noise diposits with I, >L ; further, itf s>ltb a sa.then it would still be possible to reject any noise deposit which interset two urvey lines ( assung of course, that the correlation cani b a *. froii tIn. to Una .)* Theroft *.twe hve it 0> /. t (q) S LA o if, &-3~ 3)4> SJ L< 3 1-4 -b 3~ 4 Lq Sa * I- 1 1 3 > If a wniso ' poslt is -aoceptode4hr"* to the drIflIAS sage It t, of otose, roejoete tZs z'efte: (i'O) Pj(/I.II*we1) 1o Evatuation State Probabilities .quations' (9) aM (5), ad Using the relation ?ro we Mn obtain by inspection the probability of intereepting a deposit of length I, P(it), tame: sL J P(41,1 tes f(0,Pc= PCA IAj (s") Sa lot I, 5 () satd (6), the epting a depaosit of grade P(AA) Ner theenoe s o> eqatIons " rly probability of a P(AA,) I . =0 =~ jPeA A A A, L i/ - e, 0 dA 1 .44 probabilities from equation's (8) sad (9# we han pCAeglq) ot t ,7 .4 As as re gejntL s< . Io, s L t S l3 LU:*1 'd r = *~ t Sd ar A " s- - ) - o 4+1 "" ° ndistin- A.< an Aw 'd< Am A LI loJ AM also since noise and ore deposits are shablet ( '3) E< . t,' (1 F) e &A* A i I AV$ the drilling stage is AA* governed by the total value of b deposit, The aWseptanee meChanism in thereO re it w~ald fatlitate matters if Probability density fnCtion of r in j&) , the the evaluation state, = j is substituted into j:./ and If is now derived. ) , the probability density we obtain F (/ P('l) of function of V given A : I (YlZ ) ; : V - AS - r 2& s> . It will now be useful if the following substitutions and since the various zones are made sp L , . =hu, of definition are beginning to be confusing figure 3 shows the various zones of definition. iv V v.- ..--- ... v PIV -- p>I fI .3 - ( Now >o1) = f(r la)f A A) andf Substituting for f() we have f(r) , ard completing the substitution for ,,v " I A 06k O f +4 V f ( -/ A) we have: ' r , o ovip pV~r {p 3AN V *.v*o* V, Pu'f+ v, A) -. : , ' ~B'-A ,,, I-q-I v J~ I) A. V 06)+ V A s a , V *,E Ds w Y &-V *p PI, OR -/ 1 ) -A n rp rl 3 5 (v)+:Lr V & v Ef tIX~4 ! .46 V-0 V 6 V, p'"'! myVsv I. The inte ration then gives ,4 14- W, + (*O)(. i 0- - j) . 11 f A V'-' c - we* pO -r*I f PC * -" " II(Is I- P6' 4MIf . Y4 9--v- ,,1 20 -P To avoid confusion it will be useful to s w shematioally V limits Of detinition of the v&rious along the the axis: MV 0 New sisee the natural state total nubser of deposits Is Poiteso~ disttibuted all that is needed sto the preba, bility of aoePting a depslit, represtsed bv p, to show that the probablity funtion in the evaluation state of aU bre of depo@ts is Poisson distributed with the total a (p if we 1et 4 X / #A. N.-N, A~ N.! N= ' Now at**~el Sf~~~~ ~~,~~~t AN-~VN .! C-,) sinee the summtion terms an~~~8 a~S& x9b are jast the sers P(,L) ei Jf.'9 expnsion t of e; Ad so t,-t. * e p') probability doenity fMtotinsi ve in their entirety, j( )ir, P ( interseotia, a~ept=an , , & v *-rv ), - P ( interseotion, aooeptae e ) ' and so 4 (0) d Thereforsi , rV. em I"(qrV)dr -At* S ",o,, 41 1 Nr) ,, -- The probability f=unetions of the properties of noise depo tstea be obtained in an i4ent- l fashion to the derivation of the ore dept it protbablttie tA.the the providiag that L,9 A however, if hiter oektion rate of the longer nose deposits of the A great tal ould be taken into aecount. labor tnvolved in the or deposit dterivtias an be avie4 in derlvins the nolte deposit funttons since ve oh need abilityp' PC), enad therefore only need the probthat a noie deposit will be aeoepted, this ., a) p'c ff (,i, *ffy (qj I ,A) fa A ef2.) d)a ) F(ft dA/ where a,i refer to the event of aeeeptance and interA more speeifie formulation of seetion respeatively. p' will mneeesatate making a specifle assumption of the maximu length, L. , of a noise deposit, Pmrofdtuetton State Prob bilities The production state probabilities are obviously just truneated forms of the evaluation state probabilities; (- Cru h (v)(V) o, I.v 1 vA4 11.14 6 (xo) v -IM CL- I) P(j whre - vc I voP,,.., COXPLETE EXPRFSSION OF TH: RMTURN FUNCTION The expected value of a Poisson variate is just its mean ( see Wadaworth ,1956 for a complete disoussion of the properties of the Poisson Distribution ). aecount of this,equation (1) becomes V, qjv) J,, (, E(r)-e]- EA) , ,J, On - - after substituting from equations (18) and (21). ) From equation 20, it follws that E5v)is just the submean of the distribution p) taken over the interval v.;. V V V and so VV vgu vdi Q-h) Sbstituting (23) into (22) then gives: e(r) . Vf(v)iv f4 , (tol)c - D4 Ar-.% (lJ. where j(e) is defined by equatotWs (17) and fig. 4. - -4 1$4 APPLICATION OF TRE COPLTE MODEL TO AIRBORNE PROSPCTING IN NEW BRUNSWIK En omposting this anthematieal s 0del quite a few gm as ptions had to be mde in oer simpinty to keep lox as to be frma be so * reslt the ultimte eif 1 which oapletely ~ rehensible; even so the has been derIved is seomlex enoug to defy 4itreet interin which orr this reason an aetual ase* stuy pretation the, moel . appied to a speettie problem beooaes a necessity i oxer to ob tain some idea whAt this model predlt,, of Pareameters for Case Study Estizatle As I bhae alr seare exeeiny frq f " stated data relevant t a sareity ~this th" neessitatea :aking " asf suptions which are attempts to eFroly arrive at estimates whiah are of the eorreet u te. Fr instance one reasonable estimate order of a of,a t .rs thi is the same as sayin 95% of the value of ore Mi iek a1sIs used to pay prouing tn lNew A reasma be esttate t of costs. is $ 1t00,000 I therefore 1.; ties uI' t* .Nov we do have a reference ( St~~h, 195 )wh'ith states that New larder ine has retmntly dlscontinued operatione; aooordtng to data obtaned from Nts presentO naae of ore in .Mrrate( June 14, 19956 ) the the New Latrder U Mine is approx- ltely 4 20#000#000 and so our estimates of a, seem to be reasoable. A rea estimate oftwas obtained by placing a piece of paper with a hole in it on a geologo map Brunsewk by randoly looating the paper on aomtisg the total naber of mineralizations oeourred within the hole, whose area was the of New the map and which equivalent of 100 square miles, a maxima likelehaod estimation of Ar ould be obtained ( Appendi 11 )4 turned out to be approximtely*: This estimation 2 deposits per 100 square tI.e4 lrom the 1956 we obtainerd an estimate of value oft the 4 i the Brunswick No. 12 Min. whish is lose to $ ?00,000000. No oe of reserves of other sites comes ther estates tosthis fig oe and so it seems fair to say Drieftig osts are agaitn onjeeo turd Wt if 30 holes at $ 5 000 ap eoe are noessary to evtaUate a deposit geophyss, roads, and$ 100,000 is alleed for g~ t,. then 4 250,000. Zf only 3 Gill holes ela4 are sattotent to deteraine if a deposit is noise and $ 5,i000 is allowed for roads the C6$ mi eoxphyobstes, .lai-ts anM 500o04 This . O# I arbitrartly let r:m-. rY r , ?Per anoats to sayng that on the average 9 out of 10 deposits have a 1engt less than j their maxim= possible length an 19 out of 20 deposits have a grade/length less than I believe these are stringent theL maximna grade/lengthW enoub eistimations to keep the values within reason. In estimatin C ve ca arrive at a somewhat more exact value than ftr the other parameters since most airborne surveys ru in the neighborhood of $ 20/11ne mile. If we let ",o b the length and width in ailes of the chosen area then c:r=omn'V i'vg. the data available, the two largest deposits in the Bathurst-weweastle dis- triot have a length of 1200 feet, therefore 1500 feet is probably the maximum length of any possible ore deposits in the district. If we decide to consider a 100 square Oci'. " mile area in this district, then C=7,M Computational Form of the Return Function For the purposes of computing, equation (24) can be put into a more convenient form: F (4 : where .,. t )-, (. ,, Y) a r . ,o Or -W-s , -,r ')-0 . P*) (0 ,,, , f t)) - c From figure 4 now we can see that there are ten possible return functions depending upon which interval l'.falls Fortunately,for this case !-w,,sand so unlesswe use into. very small p or MpmV&r/..,>.We now have only three diff-. I. rI ,pr erent return functions: p,>I s~, ,: Forp For m< - V)v I: t t V PL4~ k g01"V~ fn Y.aS v ,I,,,', -" (,Y , J 217 Foro wL .9vp j- O -IA +(.': *1 After inea )7 (IlulaI mn and.45 #416 ~UO,~.,-,,~w C 4I) ( all~ th ia+ -. O ., onsans as aOny faantr raiB The ottb1* it P .)m vhih up to now has bee s ely avod* i New rEeW via ft of the Prineipal *f 3t4s0 * staebn eephys ties are l ua g gr p h ite ba a m s l~ ve id enti al reats an Mber"b on ate syst 0t ar th r it seem llkely that A.otad v ea se aat."ions (13), (14.) (19), if v anlt P (q . C ,),) 5"//"( Lot us now prope :A zA. to ' (t~') bap' (S L )I It t I- A )Ifpg that 1.,iadt t can this is the $am as saying tUat the notse deposits ftollw the Sae &Mtl buttan but oly ver a wider range ad arresponds with the ru that Stphite Im **our*s In very Lmg beats. Sine sMe fttle is nmoe about th distributims of noise eposits, let us peostu38e two sodeti z. f'4, 1. a=,:O ,d7t.' o CO We may nw Oarry et the ealstuations for the two diffnre noite .sd -keeping the remainder of the parameter, at the values we have already assigned. For overall rs~rm factions" for ea lora M4m in New Brtnviek we then P sa): x , =To ,)->* ., e-.<,P,') -i.o e:.)- 0 (-is) E(k) Due to the tin to Fr) ~A ? waplezity of the J.o P) - rIP(.-. o9lets (26) the boavier of this f . P () 'C easpression o atson * 14 equa ly unmdder- stood by setually ptzg values t the return ttnotion. Piares 5 and 6 show the result obtaned by varying p, ---i --- . .. S.. . ... .. .. ... . I. 4- -- - .. .. 000 . o i i. i . .. * : r * I - I . .. f i . .o 4 . L S- , --- ... i .. ... . . i ' .... ..... . . " _L __.: ..... --.--i - -- ' I. .. . -. ---.... '- . . .. . .. .... .. . . ...... ------. ---- - ~--i ......... : - -- ..... . . - ----.-. .. -------- ------ .. --------" '1- .€ .. -------- 0a i .... .... j -.: ::::: _- . . ... ... ._ fI---~-1 - -- ... ,----o 0 Itoo', 00NCLUSIONS I s thesis has been essentially a pret inrestigatton ofa partiater eplttrn primary putoee no to establlsh a.sof irtn. of expVta&t a p probl poblem, ItS 0 the eopewat t14t e seeaMial purpose wa to .ei i at more apecialsed Ifor the oben problmt. The raen 4 a 1 epres 00idn t this thes, Lees p nnie a ramrk r the estation of othe prbles beetJe the oe svnttetl1o44 wth via tis thess an tar as the asel mel has a more wrest~le sppteabiltty this theEs Bas As. ftr el* ated i e te~d ant draft tntaah GMastia St partiewlar s proble th fit ta ti pnbablry mare (eta ne the ntl betwe any teante. it ,l fr t o, be mya n be a stMy with an tepee aof earaaty, the objetIe of Inazriolat deta, i we agree .e .ante. . pm wbta the .aelas A.eleped eato reaenbly well With experlensee, thm eartain rather lateot ee n tas ea be dramva from the wrk aoplts 0t , In reoe-ons whia 4 not have extensvet ane Spots, as eharaserte iet perhaps by dael 1, prfi slly exptentfot Wamy proSar voui have. a psitive a peoted retwn; however: there do appear to be two elati a~a I) lpa e the Ite as widely as the legth of ore depslts a ce*ept praetisallyf all despoeits d the geohstes sta e, 2) spaw* the lines very olasel tSkthe" r sA aeopt fhr triling all 4epoits iatrne et, Ina notq areas, as ehareterised perhaps by Mol .X1 the prblsm Of exploration pladmag beonos w4"acm oritesl sta if the Itns are tneeorestly spaeed no mount ts4t1ortmiatiea ill pevet a negative expeoted profit. In point of raet, the only method by which a positive expected return as be assured is by spacing the lines as wide as the maximm length of ore deposits. ?etse deductions aght peraps explain the disillusioned attitude of many people who have dealt with geophystes in the past,sor it the exploration pronsa has not ben planned earefully with reliable geolgical data, the -line spaetg might be too wide or nar e and in a noisy area this mght well mean a negative expected retour This poitts up sather conxluston them an exploration propm sho"ld not be plemed without extentive knowledsge of the area invOlved especially with respet to the costs of mini:ng the alogy, and most espeially the dtsti bution of noise deposits, Again such a eonalusion was arrived at taig before this work vas done but now we an perhaps Munderstnl just wtat types of information are necessary and why this Itformaton is se important, I fee1 that my own experienoe in working with this problem has providod some inftrmation whiah might be usefal to any other researchers in this field. I did, of course, attempt to fforalate other models besides the one presented; all the other models wwere vastly more eomplitated althouh they were also more realiste. All these other models were arried through the analytical p*ortln of their formElation, bt all these models were so intrie1te that it was Vset le to derive a coherent result from thp and It was timpossible to test if they afforded realstic answers, Judgins fro my expertenoe, I feel that future work dealing with more eoox1qt models should be done using simulation technques if specifice results are desired. APP~DIX 1 Proof that E [ Let ; J - E(N,) E(v) 4a, A(s) be the generating function of the generating funttop of funotion of the sum . 3j PIC) . fv), Bls be , and C(s) the generating Feller (1950) has shown pt that P4) C (s) (I A s) and further that if v , and Na are stoohastically independent then the riht side of (19) is just the Taylors series expansion of fore C() Z 8 e4 B(s) with s replaced by A(s), . Now if a variate X has a There- : .(X) /, , and so ation c0, khsn E senewtins Elr il T*O is. For any generating funetion of a random variable, 6r) = 1 ; therefore, I ECM,) EY. ad z+4/ 6()j and E ; -0 N) NO- 34 Maxi APPENDIX IX Likelihood Estimator of the Mean of the Poisson Distribution If a probability density f notion -fr() has a para- meter ,a. , and a if we have a group of observations of ,n . then if we.4 fine L= ., f(e), L T - t ) t ) defines the maximum likelihood estimation of in terms of S(Wadeworth.1t56). For the Poisson Distribution KN) = k and thus ) A', - J. +-1, which simplifies to -m 2wN,) N - --- NO o or IV 1S 3, BIBLIOGRAPHY A os, w. WB "The Effect of Iine Spaoingr Illustrated by Marmora, Ontario Airborne Magnetometer Control hra. ,Spaiag," G sa the Deteruemation of Opti. vwO. Altas . 4, 20, tober, 195, p. 871. pr~peets of Exploration over Large Territories,-4 X. "Meths et Appratsting EaeMno S4g Feller, Vill aa. Pry, Thornton, New Yorkt nAu Y Is . . Van W* tran , . no "Mineal Revew of New ranvsiok ' Eol. 77, no. 1, February, 194. MNt, A, N*.a L la0rthe men 87 : 3 19% so4 ts 0 atlat o6 e abridge, Nasaehusetts. SmIth, J. 0. '"ew Brunsik Reviewi." l~aal, VtL 79, no. 2, February, adiaworth, G. P, Pt, , ,...s an Copyrighted, January, 1 Note. .