At the present time, models of Earth's orbital variations are reliably accurate over the past 50 Ma (Laskar et al., 2004). Prior to this time, expectations for chaotic behavior in the solar system casts uncertainty on the modeled orbits. Within the 50 Ma timeframe, however, the models appear to be reasonably well constrained (Pälike et al., 2004) and can be used to calibrate cyclostratigraphy. Several sources of error in ATS development nonetheless require careful evaluation and monitoring, as follows.
Precision of the ATS depends on the correctness of the paleoclimate model.
Choosing the wrong insolation model for tuning climate-forced sedimentary cycles can result in precision errors of up to 12 kyrs (half-precession cycle period). This is illustrated above for hypothetical insolation curves calculated for March versus September over the past 200 Ka. Tuning to March when September insolation is the true forcing results in half-precession period tuning errors (this is an extreme case). This type of error can be minimized through examination of the obliquity-precession phasing relationship to guide selection of a best-fit insolation model.
Accuracy of the ATS depends on the correctness of dissipative Earth modeling.
Tidal dissipation through time results in a slowing of the Earth's rotation, increased ellipticity, and a reduction in the precession rate k. The figure above greatly exaggerates this latter effect, but shows that if the model for k is kept constant when true k is not, tuning errors mount with elapsed time. This example is shown for the obliquity, but the precession index is subject to an analogous effect. The indeterminancy of the effect limits the accuracy of the Astronomical Time Scale back through geological time. Modeling thus far suggests that by 20 to 25 Ma the tuning errors caused by these geodynamical effects are on order of ca. 68 kyrs for precession-tuned records and ca. 123 kyrs for obliquity-tuned records (Lourens et al., 2004).
For geologic ages previous to 50 Ma, precision and accuracy problems in the orbital solution mount rapidly. Only a few modeled planetary motions are stable enough for use as a metronome, for example, the 405-kyr orbital eccentricity cycle arising from the interaction of the secular frequencies g2-g5. Model stability studies by Laskar et al. (2004) suggest that the uncertainty of the ATS using this term alone will be at most only 0.1% at 100 Ma, and 0.2% at 250 Ma.