Travis A. O'Brien - Earth and Environmental Sciences Area

Biography

Travis O’Brien is a research scientist in the Climate and Ecosystem Sciences Division at Lawrence Berkeley National Lab and an Assistant Adjunct Professor in the Department of Land, Air and Water Resources at University of California, Davis. He currently serves as the program domain lead for the Climate & Atmosphere Process Program Domain. He also co-represented LBNL’s Earth and Environmental Sciences Area (EESA) on the LBNL Diversity & Inclusion Council from 2015-2017.

Research within Travis’ group focuses on understanding the fundamental physical processes that control weather and climate phenomena that impact human and natural systems. He specializes in utilizing a combination of numerical models, novel data analysis techniques, and fundamental theory to form and test hypotheses about what controls the physical characteristics and occurrence of weather patterns: from fog to extremes.

Research Interests

Human and natural systems are roughly in equilibrium with the statistics of how, when and where weather occurs. Changes in these statistics—of climate—can have dramatic effects on these systems: whether the changes are due to natural climate variability or due to long-term trends. Travis’ research is guided by the overarching hypothesis that there are a few key physical processes, such as convection or turbulence production by cloud-top radiative cooling, that govern the statistics of how, when and where weather occurs. Research to understand these key physical processes will therefore help us understand and predict the statistics of weather. His research largely focuses on the statistics of clouds and rain and the processes that control them using a combination of numerical models, novel data analysis techniques and fundamental theory.

Clouds: from fog to cirrus anvils

Travis’ Ph.D. dissertation, “The recent past and possible future decline of California coastal fog,” focused on running climate model simulations to understand why the occurrence of coastal fog became less frequent throughout the 20th century. A prerequisite for this research was a climate model capable of simulating coastal fog. A large part of the work involved coupling a new turbulence (boundary layer) parameterization into a regional climate model; he coupled the University of Washington turbulence parameterization into the International Centre for Theoretical Physics' regional climate model, RegCM4. The new turbulence parameterization improves the representation of the physical processes that occur at the tops of stratiform clouds, and it allows RegCM4 to develop stratocumulus clouds and coastal fog. It allowed simulations that (1) provide additional evidence that fog frequency decreased throughout the 20th century, and (2) that SST changes may not be the main driver of the trend. This effort is described in O'Brien et al. (2012, Geophys. Mod. Dev.), Giorgi et al. (2012; Clim. Res.), O'Brien et al. (2012, Clim. Dyn.), and Güttler et al. (2014, Clim. Dyn.). Travis currently serves as a community developer for RegCM.

His current and recent research on clouds includes research characterizing what controls the distribution of cloud sizes in climate models: O’Brien et al., (2013, J. Clim); and Martini et al., (2015, JAMES). This line of research uses a novel analysis of contiguous cloud regions, it develops a theory for how cloud coverage should change as the grid-spacing of numerical atmospheric models decreases, and it shows that the representation of stratiform cloud processes are key for determining whether atmospheric models obey this theory. Travis is also starting some new research into the frequency of coastal fog events and how much water the fog transports (part of the multi-institutional Summen Project).

Rain from local to global scales

Much of Travis’ current work is focused on what governs the probability of extreme rainfall: this work is part of the Calibrated And Systematic Characterization, Attribution and Detection of Extremes Scientific Focus Area (CASCADE SFA), which he co-leads with Bill Collins. In this work, Travis and colleagues have developed and evaluated a theory for why extreme precipitation depends on grid resolution in atmospheric models. Rauscher et al. (2015, J. Clim) lay out theoretical arguments for why extremes would depend on resolution, O’Brien et al. (2016, JAMES) demonstrate that this theory explains the observed increase in intensity of extremes in a well-controlled set of hindcast simulations, and Donner et al. (2016, GMD) argue that this theory could be leveraged to develop simple ‘scale aware’ parameterizations of updraft. Ongoing research in this topic focuses on what governs joint probability of intensity and duration of rain storms.

Community Impact

Community impact is a key component of Travis’ work at LBNL and UC Davis. He has co-convened six sessions on fog each year at the AGU Fall Meeting since 2012; these sessions have served as a center for the coastal fog research community and have catalyzed new research projects (including the Summen Project). He has also been a co-organizer for three recent workshops: the 2015 DOE-RGCM Team Leads Meeting, the 2015 DOE/NOAA Workshop on High-Resolution Coupling and Initialization to Improve Predictability and Predictions in Climate Models, and the 2016 DOE-RGCM PI Meeting.
In addition to community leadership, Travis has produced tools and datasets that are freely available to the broader community. As part of his work in CASCADE, Travis has produced a huge ensemble of hindcast simulations (generated the InitiaLIzed-ensemble Analysis and Development [ILIAD] hindcast framework described by O’Brien et al. 2016, JAMES) that are publicly available at NERSC for use by other researchers. Additionally, in his research on the resolution dependence of precipitation, Travis developed a fast, robust and multidimensional kernel density estimation tool, described by O’Brien et al. (2014, CSDA) and O’Brien et al. (2016, CSDA). The tool, fastKDE, is available for use in Python by the general scientific community and is easily installed using pip (`pip install fastkde`).