ABSTRACT Predictions of global climate change are based on large computer-simulation models that are “history-matched” to weather records compiled from the early nineteenth century onward. Climate-change model forecasts would be more convincing if they were based on the natural records of the Holocene (≈10,000 years) and were capable of simulating climate characteristics of this epoch. Temperature records estimated from δ 18 O measurements on ice cores from the Greenland ice cap and the Antarctic could be used to develop models based on geochronological data rather than historically brief weather records. The 20-year average record of δ 18 O values from the Greenland Ice Sheet Project 2 (GISP2) ice core exhibits a long-term trend of declining temperatures over most of the Holocene, except during the last 100 years, when temperatures have increased—a change widely blamed on carbon-dioxide (CO 2 ) emissions from fossil fuels. However, the range in temperatures since the start of the industrial age is typical for the Holocene, and the current rate of increase in temperatures is unusual but not unprecedented. Past periods of consistently increasing (or decreasing) temperatures have not persisted much longer than the current interval, so temperature trends may well reverse in the near future. There are distinct cyclic patterns in temperatures recorded in the GISP2 ice core, including a pronounced sawtoothed 560-year sequence of relatively abrupt change followed by a gradual reversal. The present trend may be the initial phase of such a pattern. In summary, the present climate does not appear significantly different from the past climate at times prior to industrialization.

The digital computer is undeniably an asset in the examination of many classes of geologic problems. Unfortunately, it is ill suited for handling pictorial information which constitutes a large percentage of geologic data. For certain kinds of problems, the inherent physical properties of optical lenses can be used to perform analyses that are impractical using a digital approach. For example, in a current study of pore structure in reservoir rocks, the pore pattern of an area 24 × 24 millimeters on a thin section was digitized, yielding more than one million data points. Spectral analysis was used to determine the relative contributions of spatial frequencies to the total porosity, but even with the Fast Fourier Transform, a two-dimensional spectral analysis of a single thin section is very expensive even on a large computer. In contrast, a proper optical lens system will produce a Fourier transform and map the power spectrum on film in a few seconds. A digital approach is more expensive by three or four orders of magnitude. Optical-processing methods are especially well suited for study of radar imagery air photographs and gross fabric patterns, as well as microscopic textures in rocks.