USGS visual identity mark and link to main Web site

USGS Open-File Report 94-023

Model-Model Comparisons And Data-Model Comparisons: Considerations For The PRISM Paleoclimate Study

Lisa C. Sloan
University of California-Santa Cruz, Santa Cruz, CA 94064
Because plans in the PRISM collaboration call for Pliocene climate simulations to be produced by two different climate models (GISS and NCAR), a comparison of the two models is necessary at the start. Both models are three-dimensional general circulation models (GCMs) of similar design, but there are important differences in specific climatic processes parameterized by the models, and in spatial resolution, among other factors. Therefore, before Pliocene results from the two GCMs are compared in this project, the models must be compared at a present day baseline climate. The strengths and weaknesses of each GCM should be documented, and differences in model performance under present day conditions between the models and observations must be considered before the global Pliocene results are examined. Some of the model characteristics have been documented by other model users, but we need to compare the model versions that will actually be used at each phase of the PRISM project. We should also keep in mind how the individual model parameterizations relate to factors thought to be important to Pliocene climate.

A hierarchy of comparisons of the model results should be carried out. First-order comparison will be of global mean annual values, at least for temperature, precipitation, top of the atmosphere energy elements, and general model sensitivity to forcing (e.g., CO2 doubling). Second-order comparisons of zonal averages will include comparison of model output fields of temperature (annual and seasonal), precipitation (annual and seasonal), sea ice, snow, zonal wind speed at near and mid troposphere levels, vertical velocity at midtroposphere level, and energy components. Third-order comparisons should include global (map view) distributions of minimum, mean annual, and annual amplitude of surface temperature, mean annual and seasonal precipitation, soil moisture, runoff, and total moisture (precipitation-evaporation). Statistical differences for these fields may prove useful. Other climate parameters that may be of specific relevance to Pliocene climate in general and the PRISM data base in particular should also be examined for these present day cases.

Comparisons between model results and data are an integral part of the long-term goals of PRISM. Comparisons at various stages of the project will serve as the means of evaluating both model results and the synoptic mapping efforts. These comparisons will also be used to plan for future interactions of climate experiments. There are several factors to consider when comparing model results to proxy data interpretations. First, model spatial and temporal resolutions will almost certainly not correspond to the resolutions reflected in the paleoclimate data. This must somehow be taken into account in comparisons. Second, the GCMs parameterize many processes instead of explicitly modeling them, and this may affect the results in relation to data interpretations of climatic factors. Third, careful consideration must be given to the question of what the proxy data have actually recorded; do pollen or macrofaunal assemblages most closely reflect mean annual temperature, growing season temperature, or some other quantity(ies)? This is critical when different types of proxy climate information are related to each other and are compared to climate model results. Last, multiple independent proxy indicators for a given climate parameter are useful to obtain because they yield multiple signals for a given system. This again is useful information for understanding the Pliocene climate system and its possible forcing factors.

Proxy climate data can be compared to model results on the basis of quantitative or qualitative information. Qualitative information includes wet/dry and hot/cold indicators which provide a first-order test of model results. Quantitative information includes minimum temperature, mean annual temperature, mean annual temperature range, mean annual precipitation, and seasonal precipitation, can be used as a second-order test of model results.


This page is <https://pubs.usgs.gov/openfile/of94-023/17_Sloan.html>
Maintained by Eastern Publications Group Web Team
Last modified 28-Feb-2001