1. An oft-quoted saying goes “It is difficult to make predictions, especially about the future”. This is especially true of a system as complex as Earth’s climate, though we can thankfully rely on well-established conservation laws to make quantitative statements about the evolution of climate over space and time. Much of our knowledge of climate dynamics is thus encapsulated into global climate models (GCMs), which mathematically encode the physical, chemical, and biological processes that govern the evolution of Earth's outer envelopes.  How much should we believe their predictions? This is a harder question than it looks, as the instrumental record (starting around 1850) is too short to reliably test predictions made a century, or even a few decades, ahead of time. Only the longer paleoclimate record can help in this validation effort. Thus, if we want to know the future, we must know the past.

  2. Our research uses paleoclimate information to better understand Earth's low-frequency climate variability, the bassline of climate. We focus on the last 2,000 years of Earth's history (the Common Era), for two reasons. On one hand, it is long enough to provide adequate sampling of the slow modes of natural climate variability. On the other hand, it is short enough that (i) the mean climate state was close to today's; (ii) many high-resolution paleoclimate records are available, and (iii) its climate can now be simulated using state-of-the-art GCMs. This conjunction allows for an unprecedented window into low-frequency climate dynamics through the confrontation of climate models and paleoclimate observations.

  1. This, too, is more complicated than it seems. As the figure above shows, the output of climate models and our observations of past climates (proxies, be they from ice cores, tree rings, cave deposits, corals, or sediment cores) are expressed in different languages: GCMs speak the language of physics (temperature, pressure, vorticity, entropy), while paleoclimate observations typically speak the language of geochemistry (e.g. the abundance of water isotopes, trace metals or organic molecules). Thus, in order to use paleoclimate observations to constrain climate models, one must erase this language barrier. Our lab does this in two ways:

  2. 1.Inverse Modeling: building statistical representations of how climate affected proxies and inverting this relationship to back out which climate generated the observed proxy values (lower arrow on the figure above). We use both frequentist and Bayesian methods to do that. Recent examples are included here, here and here [more to come].

  1. 2.Forward Modeling: building dynamical models of the Earth’s climate (e.g. explicitly representing the physics of water isotopes in the atmosphere and ocean) and developing process-based models of how proxies incorporate climate information (upper arrow on the figure above). Examples of this work are available here and here. [more to come].

  1. It is our belief that some truth lies somewhere in between, though we are still searching for it.  For more details, please visit our publications page.