The post-1998 slowdown of global average surface air temperature (T) warming has become commonly known as the "global warming hiatus". During the same period, regional precipitation patterns (P) have been markedly different from the decades before and, further, there has been an apparent decrease in globally averaged precipitation. These trends are not reflects in the global coupled climate models used for the fifth  climate assessment of the Inter-governmental Panel on Climate Chang (IPCC). At the time this proposal was written, it was not clear if this discrepancy indicates that these model are in error or whether it is a result of a strong natural climate fluctuation that is temporarily interfering with the impact of the still increasing greenhouse gas concentration. Moreover, there is mounting evidence suggesting that the warming hiatus is tied to persistent cooling of the ocean surface in the eastern equatorial Pacific, and that the anomalously cool sea surface temperatures are in this case driving the observed global climate conditions. Here too, there is great debate on whether these tropical conditions arise purely from internal climate variability or are externally forced by anthropogenic aerosol emissions or recent weak volcanic eruptions.

In order to differentiate between natural internal variability and responses to external forcing, we will apply the well-established statistical framework of climate change Detection and Attribution (D&A) to changes in P, T, and in the pattern of the precipitation response (that is the rate of change in precipitation with respect to changes in surface temperatures). There are several advantages to the D&A-based approach. First, detection and attribution provides an excellent framework for model evaluation. If the hiatus, and associated hydroclimate change pattern (or fingerprints), are attributable to internal variability alone then models would be expected to reproduce only the magnitude, but not the timing, of the trends (that is their timing of producing a similar change will not necessarily coincide with the actual, observed change). Conversely, if they are attributable to external forcing, then the inability of the models to reproduce them would indicate model errors in forcing or response. Thus, approaching the hiatus period as a D&A problem can aid in the development of physically defensible metrics for model performance. Second, the fingerprinting techniques that are used in D&A analysis differentiate between the climate responses to different forcing terms; that is, each forcing component, Greenhouse gas increase or changes in atmospheric concentration of industrial aerosols, tends to have a different T, P, and precipitation response patters. Thus the spatial patterns themselves yield insight into the physical processes that affect the response. In the proposed work, we will estimate the changes in the latitudinal and seasonal distribution of rainfall that are expected in response to each external forcing, and interpret those changes in terms of thermodynamically and dynamically induced precipitation changes. Thus, while this project is focused on recent climate change, the multivariate fingerprints developed for it will be more generally applicable.

In addition to applying D&A methods to climate model simulations, we will conduct a series of model simulations with the most advanced version of the coupled climate models of the National Center of Atmospheric Research (CESM- CAM5). These will be aimed at estimating the relative roles of radiative forcings and natural variability in the recent T and P trends. Our control mode integrations will be forced by the time history of observed sea surface temperatures (SST), sea ice, and trace gases, and based on similar past experiments are expected to reproduce the salient aspects of observed change in T and P. Additional simulations will be carried out to estimate the role of each forcing separately. For those we will use the changes in SST and sea ice induced by each radiative forcing separately as identified from "single forcing" experiments that were done as part of the IPCC model runs. These experiments, in combination with those with actual historical SST forcing, will assess the extent to which changes in P and T over recent decades were driven by natural variability and by the various radiative forcings. These estimates will be compared to those derived from the D&A work to provide increased confidence in the results.

By addressing the question of whether model error in external forcing and response or internal variability is responsible for decadal differences between simulations and observations, the proposed project furthers the goals of DOE’s Regional and Global Climate Modeling Program to assess and increase the reliability of climate change projections.