The calendar effect in PMIP4 time-slice and transient simulations: overall impact and strategies for data analysis

Session: Cross-cutting Group 2 (Paleovar, Past to future, Data assimilation)
Author: Patrick J. Bartlein / bartlein@uoregon.edu / University of Oregon
Co-author: Sarah L. Shafer, U.S. Geological Survey;

Abstract:
The “calendar effect” is the common expression for the impact that the known changes in the length of months or seasons over time, related to changes in the eccentricity of Earth’s orbit and precession, have on summarization of model output. Even if daily data are available, the calendar effect must still be considered when summarizing data by months or seasons, or when calculating climatic indices such as the temperature of the warmest or coldest month—values that are required for comparisons with paleoclimatic observations. The impact arises not only from the changing length of months or seasons, but more importantly, from advancement or delay in the starting and ending dates of months or seasons relative to the solstices. The impact of the calendar effect is large and spatially variable, and can produce apparent spatial patterns that might otherwise be interpreted as evidence of, for example, high-latitude amplification of temperature changes, continental/marine temperature contrasts, or variations in strength of the global monsoon. Calendar effects must also be considered in the analysis of transient climate-model simulations (even if data are available on the daily time step). Time series of data aggregated using a fixed modern calendar as opposed to an appropriately changing one can differ not only in the shape of long-term trends, but also in the timing of Holocene “thermal maxima” by several thousand years, depending on the time of year. There are a number of approaches for adjusting monthly data that were averaged using present-day calendar definitions to a “paleo calendar”. A simple one involves a) determining the appropriate fixed-angular month lengths for a paleo experiment (e.g., Kutzbach and Gallimore, 1988, JGR 98:803-821), b) interpolating the data to a daily time step using a mean-preserving interpolation method (e.g., Epstein, 1991, J. Climate 4:365-368) or using archived daily data directly, and then c) averaging or accumulating the interpolated daily data using the appropriate paleo month starting and ending days (i.e., month lengths). We present examples of the calendar effect and discuss their implications for interpretation of paleoclimate data.