Does the use of reanalysis-based modern climate datasets improve paleoclimate reconstructions?

English Abstract:

The use of different modern climate data sets has a significant impact on a continental-scale pollen-based reconstruction of mean July temperature (TJUL) over the last 2,000 years for North America. We use modern data from climate stations, data physically modeled from climate stations and several reanalysis products to calibrate the reconstructions. Our results show that the use of reanalysis products produces warmer and smoother reconstructions as compared to using station-based data sets. The reconstructed TJUL anomaly (computed from the average anomaly of all sites in North America) is more variable during the period of 1050--1550 CE due to a high latitude cold-bias in the modern TJUL data used. Since the anomaly is reconstructed minus modern TJUL the cold-bias in the modern data results in a warm-biased anomaly. The ultra-high resolution WorldClim gridded dataset may only useful if the modern pollen sites have at least the same spatial precision as the grid resolution and if some type of statistical downscaling approach is used to improve the climatology of each 1 km grid box to match the climatology of the nearest modern pollen site.

Due to the limitations of reanalysis products and high resolution grids that are physically modeled from stations, we prefer the use of a station-based data such as the lapse-rate corrected University of East Anglia Climate Research Unit (CRU-based) Whitmore modern climate data set for North American paleoclimate reconstructions from pollen.