the other end. It is possible, however, to obtain fairly close agreement over sectors"'
  from ±80° to ±180°, as seen in Figure 9. Since for Atlas HAS there are few, if any,
  significant population centers in the launch area outside these sectors (i.e., within ±80°
  of the flight line), failure of the curves to match closely near the flight line is of little
. consequence. If a better data match is considered desirable for computing risks to
  population centers within ±80° of the flight line (e.g., ships), either a different A can be
  selected for use with B = 1,000 or other values of A and B can be derived. If only a
  single value of B is used, no matter what the value, a good match between theoretical
  and simulated data is not possible over the entire 180° sector for various breakup qa.'s.

 Before becoming too concerned about lack of a data match between 0° and 80°, it
 should be remembered that many types of Mode-5 responses cannot be simulated, so
 that the malfunction-tum impact distributions plotted in Figure 9 are only a subset of
 all possible Mode-5 impacts. Based on twelve Mode-5 failure responses for. which
 impact data are available, it is believed that inclusion of the ''non-simulatable" Mode-5
 responses would considerably improve the match in the sector from ±10° to ±80°.
 Another mitigating factor is that risks near the flight line are totally dominated by
 Mode-4 failure responses.
 To see how data matching is affected by selecting widely differing values of B, the
 theoretical Mode-5 impact distributions were computed for B =50,000, 100,000, 500,000,
 and 5,000,000. Best-fit values for A were again determined by trial and error. Results
 are shown in Figure 10 through Figure 13 along with the same impact distributions for
 random-attitude turns plotted in Figure 9.




 "' For other values of B and qa, close agreement is possible from ±60° to ±180°.


 9/10/96                                            43                                     RT!