Statistical Uncertainty in Paleoclimate Proxy Reconstructions.

Area of Science:

• Paleoclimatology
• Statistical modeling
• Environmental science

Background:

• Quantitative environmental reconstructions rely on proxy data, especially for periods predating instrumental records.
• Underestimated uncertainties in proxy reconstructions can cause conflicts between different proxies and between proxy data and climate models.
• Accurate uncertainty quantification is essential for reliable paleoclimate interpretations.

Purpose of the Study:

• To present a simple, robust, and generalizable method for quantifying uncertainty in proxy reconstructions.
• To identify key factors controlling the magnitude of uncertainty in proxy-based environmental reconstructions.
• To compare the proposed uncertainty estimation method with those from Bayesian, nonparametric, and fiducial statistical frameworks.

Main Methods:

• Utilized ordinary least squares linear regression as a foundational example for uncertainty quantification.
• Developed a method to quantify uncertainties inherent in proxy-based environmental reconstructions.
• Compared the developed method's uncertainty estimates against established statistical approaches.

Main Results:

• The proposed method provides a straightforward and reliable way to estimate uncertainty in proxy reconstructions.
• Unexplained variance in calibration data was identified as a primary control on reconstruction uncertainty.
• The simple method's uncertainty estimates were found to be comparable to those from more complex statistical frameworks.

Conclusions:

• The unexplained variance in calibration data must be explicitly included in the uncertainty assessment of proxy reconstructions.
• Future research should focus on maximizing the explained variance in calibration datasets to reduce reconstruction uncertainties.
• A 'data-forward' approach is recommended, separating proxy data presentation from environmental interpretations.