nav2.gif (5736 bytes) title.gif (3369 bytes)
Calibration Spline

The evaluation of both Cariaco Basin varve and radiocarbon chronologies provides independent evidence that the chronologies are accurate and can be used as an alternative data set for 14C calibration during the period of deglaciation. In order to provide continuous calibration coverage, an interpolation spline was fitted to the Cariaco Basin data set (Figures 3 and 4). This spline was chosen over other possibilities (for instance, cubic spline, linear interpolation) for several reasons. First, the spline approximates a linear interpolation between points, which is a simple interpretation and avoids the risk common to cubic splines of creating structure beyond the resolution of the data. Second, the spline has the advantage over a linear interpolation of providing a continuous curve which passes smoothly through the data points. The interpolation spline shows excellent agreement with the German pine data (r = 0.99) and preserves detailed century-scale variability, for example during the Preboreal period, around 9.6 14C kyr BP (Figure 3).

The interpolation spline was used to create a curve through the entire Cariaco Basin data set to extend continuous 14C calibration from 9.0 to 12.7 14C kr BP (Figure 4). This represents an objective treatment of the calibration in that no Cariaco Basin data points are excluded, and the same interpolation procedure used to match the curve to tree rings is used to extend the curve further back in time. The full Cariaco Basin calibration spline is plotted together with paired 14C-U/Th dates from Atlantic and Pacific corals, and agrees in general with corals throughout most of the record (Figure 4). However, the data sets do show some large differences, particularly during the Preboreal and Younger Dryas periods. The Cariaco Basin data show less scatter around the tree-ring data than corals, suggesting greater reliability during earlier periods as well. The Cariaco Basin data also provide greater resolution and bridge the numerous gaps in the coral data. The new Cariaco curve resolves detailed changes prior to the 9.6 14C kyr BP plateau, including the precise timing of the beginning and end of a long sloping ‘plateau’ with superimposed century-scale structure during the Younger Dryas. In addition, there are plateaus at 11.7 and 11.4 14C kyr BP, and the possibility of a brief plateau or reversal immediately preceding the Younger Dryas at 11.1 14C kyr BP.

 

Identifying features such as plateaus in the radiocarbon calibration curve, and determining their relationships to abrupt climatic events occurring at the same time, is crucial for a wide range of paleoenvironmental studies. Calculating rates of change or sedimentation rates in 14C-dated sediment cores depends greatly on whether, and how, the 14C dates are calibrated. For example, a sediment age-depth curve based on uncalibrated radiocarbon dates will introduce error into rate-of-change and proxy flux calculations. Using uncalibrated 14C ages to calculate sedimentation rates creates large anomalies, artificially increasing values during plateaus when the radiocarbon ‘clock’ is running slower than calendar time, and decreasing values when radiocarbon is running faster.Uncalibrated 14C chronologies can result in substantial, abrupt changes in sedimentation rate (and rate of change) calculations that coincide with, but do not necessarily relate to, paleoclimate change. Ironically, this problem appears to be compounded by high-resolution sampling for 14C dating, as a limited number of 14C dates will tend to smooth over the sharp bends in the calibration curve and thus produce a smaller, smoothed anomaly during the Younger Dryas.

This new radiocarbon calibration data set from the Cariaco Basin is available for use as a higher-resolution alternative to curves based on coral 14C-U/Th dates during deglaciation (11.5-14.5 cal kyr BP), and as a complement to longer curves based on corals that extend back to 30 cal kyr BP. The Cariaco Basin spline data are provided at decade resolution for ease of calculation on the part of the user, because 14C dates are typically reported to the nearest decade. However, it should be noted that the resolution of the actual calibration data set is approximately one date per 100 years and does not resolve decade-scale changes in D14C that may have occurred.