Reconstructing the surface temperature fields of the mid-Pliocene Warm Period using climate models and a variety of climate proxy data sets.

We discuss the construction of a global climatological reconstruction of the mid-Pliocene Warm Period, which is the focus of the current PlioMIP (around 3.2 Ma BP). Our method uses an ensemble Kalman smoother to blend fields of model output, generated here by two generations of the PlioMIP project, with sparse proxy-based estimates of temperature. The methodology was previously presented by Annan et al 2022 in application to the Last Glacial Maximum. In this presentation we contrast results obtained from two different data compilations. Firstly we use the PRISM proxy estimates for sea surface temperature from an interval 15ka either side of MIS KM5c (Foley and Dowsett 2019, Haywood et al 2020), that is 3.205+-0.015Ma. This contains 34 data points on the 5x5 degree grid that we use for our analysis. We contrast this with analyses performed in the PlioVAR project (McClymont et al 2019) covering a slightly narrower interval of 3.205+-0.01Ma. These analyses are based on a different age model and multiple calibrations of UK37 and Mg/Ca. The full data set contains 31 gridded data points.

Using the PRISM data, our estimate of the global temperature anomaly is considerably warmer than most previous estimates, suggesting a significant discrepancy between the models and the data. Further calculations suggest that the PlioVAR data show more consistency with the models, but uncertainty remains high. We discuss the implications of this for the creation of reliable climate reconstructions.