FISHERY BULLETIN: VOL. 74, NO. 3 



culated from overflight data from the time period 

 for which the estimate is to be made. 



Estimator II 



Days on grounds reported to ICNAF (g) were 

 correlated with observed days on grounds {g') and 

 fleet size (V) as established from aerial surveil- 

 lance data to estimate the probability of a day on 

 grounds that was not sighted [PiG/N)] for any 

 desired time period (Af): 



P{G/N) = ig-g')^ {{VM)-g'). (2.1) 



(See Appendix for the derivation of this 

 probability and for the resulting estimator for 

 days fished.) In addition, the relation between 

 days fished (J^ and days on grounds {g) may also be 

 established from reported effort: 



K=f^g. 



(2.2) 



Fishing effort may then be estimated for some 

 future time period from overflight data by as- 

 suming the A" and Pr{G/N) previously established: 



f=K[P{G/N){V-M-g') + g']. (2.3) 



Estimator III 



The probability of a day fished if not observed 

 [P{F/N)] may be computed for any time period 

 {\t) for which reported days fished (/), the number 

 of vessels present {V, as determined from vessel 

 identification numbers observed on overflights), 

 observed days on grounds (g'), and observed days 

 fished if') are available: 



P{F/N) = (f-f)/{{Y-M)-g'). 



(3.1) 



(See Appendix for derivation of this probability 

 and of the resulting estimator.) If this computed 

 probability is assumed for some future time period 

 for which reported effort is not available, days 

 fished for that time period may be estimated: 



f=P{F/N)-{{lt-V)-g')+f'. 



(3.2) 



In order to develop this algorithm, it was assumed 

 that a vessel did not fish at all during the day it was 

 sighted if it was observed in the nonfishing mode. 

 Since surveillance flights were usually completed 

 before afternoon, it is possible, as noted earlier, 



that evidence of fishing was not observable if the 

 vessel did not fish until late in the day so that the 

 above assumption may have been violated in some 

 cases. If this occurred the P{F/N) is incorrectly 

 calculated, a situation having no effect on the 

 estimates of days fished if such inaccuracies are 

 constant from one time period to another. Since 

 vessels are usually engaged in fishing operations 

 whenever sea conditions permit, such inaccuracies 

 can occur only during days when sea conditions 

 disallow fishing during morning hours (when 

 surveillance flights usually occur) and permit 

 fishing later during the day. If the frequency of 

 such weather conditions are assumed to be con- 

 stant the magnitude of these inaccuracies may 

 also be expected to be unvarying. 



RESULTS 



Reported effort and aerial observation data 

 from 1969-73 (Table 1) were used in the various 

 equations to compute i?(estimator I), P{G/N){es- 

 timator II), /^(estimator II), and P(F/AO(estima- 

 tor III). The number of surveillance flights {A, 

 Equation 1.1) is required to calculate R. The 

 numbers of flights for 1969-73 were 64, 66, 91, 105, 

 and 109, respectively. The P{F/N) and R were 

 computed for each gear-country category, for each 

 country, and for all stern trawlers and all side 

 trawlers for each year, 1969-73 (Table 2). Since 

 days on grounds were not consistently reported 

 except by the German Democratic Republic (GDR) 

 and in fact were never reported by some countries, 

 K and P{G/N) could not be calculated in many 

 cases. 



The variables R, P{F/N), and PiG/N) exhibit no 

 trends of increase or decline through the years 

 examined; however, these values varied, at times 

 substantially, from year to year. Therefore, in 

 order to decrease estimation error, these variables 

 were averaged whenever possible over years 

 preceding the year for which the estimate was 

 made. The average value was then used to make 

 the estimate. These variables for 1969-72 were 

 averaged to make the 1973 estimates; 1969-71 were 

 averaged to make the 1972 estimates; 1969 and 

 1970 were averaged to make the 1971 estimates; 

 and the 1969 values were used to make the 1970 



estimates. 



As stated above, days on grounds were in- 

 frequently reported so that such sequential aver- 

 aging of the P(G/AO(estimator II) was not possible 

 except in the case of the GDR. The Union of Soviet 



506 



