McKinnon group

Internal variability and climate change

Observed weather and climate are a result of the complex internal variability of the earth system and anthropogenically-induced climate change. Depending on the question of interest, one or the other of these may dominate. For example, multidecadal trends in global-average surface temperature provide a clear indication of anthropogenic influence, and ocean heat content steadily climbs upward as we add more heat to the earth system. In contrast, at the local and regional scale, it becomes harder to clearly extract the climate change signal from internal variability. The ability to do so is critical for planning for the future in a nonstationary climate, as well as for proper evaluation of climate models.

In recent work, we have focused on developing an observationally-constrained statistical model for the spatiotemporal structure of interannual variability. The output of the model, termed the Observational Large Ensemble (Obs-LE), can be used to quantify the envelope of potentially irreducible uncertainty that emerges due to the random sampling of “climate noise”. On regional scales, this uncertainty can be of comparable magnitude to the forced trend in temperature, and much greater for precipitation.

The other (much more challenging) piece of the puzzle is understanding whether and how internal variability is changing due to anthropogenic influence. During winter, we have found evidence of a decrease in within-season day-to-day temperature variability in North America, while changes in the shape of summer temperature distributions are smaller. Determining whether or not these trends are due to anthropogenic influence requires understanding their physical origin. The decrease in winter variability appears related to polar amplification, but many other questions remain: Are land surface feedbacks changing summer temperature distributions? To what extent are we seeing changes in atmospheric variability, and how does that imprint on surface temperature variability?

Related papers:

McKinnon, K.A. and C. Deser. Internal variability and regional climate trends in an Observational Large Ensemble. Journal of Climate, doi:10.1175/JCLI-D-17-0901.1.

McKinnon, K.A., A. Poppick, E. Dunn-Sigouin, and C. Deser. An 'Observational Large Ensemble' to compare observed and modeled temperature trend uncertainty due to internal variability. Journal of Climate, doi:10.1175/JCLI-D-16-0905.1.

Rhines, A., K.A. McKinnon, M.P. Tingley, and P. Huybers. Seasonally Resolved Distributional Trends of North American Temperatures Show Contraction of Winter Variability. Journal of Climate, doi:10.1175/JCLI-D-16-0363.1.

McKinnon, K.A., A. Rhines, M.P. Tingley, and P. Huybers, 2016: The changing shape of Northern Hemisphere summer temperature distributions. Journal of Geophysical Research: Atmospheres 121 (15), 8849-8868, doi:10.1002/2016JD025292.

Heat waves, droughts, and compound extremes

As temperatures warm, the probability of high-impact heat waves increase, which have an outsize influence on human health, crop productivity, and the stability of our infrastructure. Indeed, the recent 4th National Climate Assessment found that, under a high-warming scenario (RCP8.5), the annual cost of heat-related mortality alone would be $141 billion by 2090, and severe drought years such as 2012 have caused tens of billions dollars of losses. Thus, it is increasingly important to better predict when these extremes will occur with enough lead time that cities, farmers, and individuals can prepare. In some cases here are identifiable precursors in the slowly-varying components of the earth system such as sea surface temperatures or the moisture content of the land surface. Having identified and validated one such relationship for Eastern United States heat waves, the next step is to develop methodologies that can be applied more generally to pull out statistically causal links within the climate system, and to test those links with dynamical modeling.

In many cases, temperature does not tell the full story. Drought result from a combination of warm temperatures and a lack of moisture availability with impacts on agriculture and wildfire risk, and the human body’s ability to cool itself is linked to both temperature and humidity. To better understand these compound extremes, we are developing methods to describe their covariance and temporal evolution, and to identify both physical precursors and downstream impacts.

Related papers:

McKinnon, K.A., A. Rhines, M.P. Tingley, and P. Huybers, 2016: Long-lead predictions of eastern United States hot days from Pacific sea surface temperatures. Nature Geoscience 9, 389–394, doi:10.1038/ngeo2687.

The seasonal cycle

The seasonal cycle is the dominant mode of temperature variability at the global scale, with temperature swings between seasons larger than those between glacial and interglacial periods. Because the seasonal cycle results from large external forcing and is observed on an annual basis, it provides us with an observational laboratory with an unusually large signal-to-noise ratio for the response of the earth system to heating.

We have previously developed a simplified advection-diffusion model for the seasonal cycle in near-surface air temperature, demonstrating that the majority of seasonal variability can be explained by the mean circulation and the geometry of the land masses. The seasonally-constrained model can also be used to predict the spatial structure of interannual variability as well as that of multi-decadal changes in temperature. There remain myriad other ways to use the seasonal cycle as a laboratory to answer key questions in climate change science. I am particularly interested in the influence of a changing land-ocean surface temperature contrast on the atmospheric circulation, and the development and decay of seasonal drought.

Related papers:

McKinnon, K.A., A.R. Stine, and P. Huybers, 2013: The spatial structure of the annual cycle in surface temperature: Amplitude, phase, and Lagrangian history. Journal of Climate 26 (20), 7852-7862, doi:10.1175/JCLI-D-13-00021.1.

McKinnon, K.A. and P. Huybers, 2014: On using the seasonal cycle to interpret extratropical temperature changes since 1950. Geophysical Research Letters 41 (13), 4676-4684, doi:10.1002/2014GL060404.

McKinnon, K.A. and P. Huybers, 2016: Seasonal constraints on inferred planetary heat content. Geophysical Research Letters, doi:10.1002/2016GL071055.

Mountain glaciers

Mountain glaciers respond to changes in temperature and precipitation, and can provide a record of past climate changes in their moraines. Interpreting these moraine records, however, requires a careful consideration of the local environment of the glacier, including the coupling between the ice and solid earth via erosion, transport, and deposition of sediment, which I have previously explored in the context of the Last Glacial Maximum Pukaki Glacier in New Zealand. Mountain glaciers are also responding to modern-day climate change, making monitoring increasingly important. During (Northern Hemisphere) spring 2014, I participated in a small field campaign using drones to take thousands of aerial photos of Tasman Glacier, which can be stitched together to make a three-dimensional digital elevation model of the glacier for use in velocity modeling and long-term volume monitoring.

Related papers:

McKinnon, K.A., A.N. Mackintosh, B.M. Anderson, and D.J.A. Barrell, 2012: The influence of sub-glacial bed evolution on ice extent: a model-based evaluation of the Last Glacial Maximum Pukaki glacier, New Zealand. Quaternary Science Reviews 57 46-57, doi:10.1016/j.quascirev.2012.10.002.

Model philosophy

The climate system is extremely complex, with important processes spanning a wide range of spatial and temporal scales. Creating models that are also complex, however, leads to challenges in interpreting both the actual climate system and the climate models. Model evaluation through established frameworks in information theory and machine learning can assist in choosing models that maximize information content and avoid overly high variance.