Since 1976 AP-8 has been the standard model for making predictions of
fluxes of energetic radiation-belt protons. It is based on satellite d
ata acquired up to 1970. The model uses B/B-0, L coordinates and store
s omnidirectional fluxes (f); B-0 is the equatorial field strength. Th
e maximum B/B-0 for each L represents the atmospheric cut-off (ACO) wh
ere the flux gradient is large. The standard format aims to provide go
od resolution close to ACO. We report on problems with the standard in
terpolation method which give erroneous results at low altitude. When
a grid of geographical points at a specific altitude are transformed t
o B,L coordinates, and used as input to AP-8 via the standard software
, the resulting f contours exhibit an erroneous ''rippling'' structure
due to interpolation problems. The low-altitude South Atlantic anomal
y corresponds to L values of similar to 1.5 and Space Station and most
shuttle missions have inclinations < 30% where predictions are influe
nced by the lowest values of L. A coordinate system providing better r
esolution has the B/B-0 range normalized: phi = sin(-1)[(B - B-0)/(B-m
ax - B-0)], where B-max is the field strength at the ACO. At the geoma
gnetic equator phi = 0, while phi = 90 degrees at the ACO-the rapid va
riations near B-max are spread in angle. Contours produced by interpol
ation in phi,L space are much smoother and ripples are removed. We ide
ntify specific problem values and the need for at least one further L
value in the models at low L. Overall flux values with the improved me
thod are higher. It may be possible to introduce a solar-activity depe
ndence of fluxes by computing a solar-activity-dependent B-max. Copyri
ght (C) 1996 Elsevier Science Ltd.