Skip to content

A brief description of my major accomplishments at GFDL


Development of GFDL’s global high resolution atmospheric model HiRAM and studies of hurricane-climate connections

I was the lead developer of GFDL’s HiRAM, which has led to a major advancement in GFDL’s capability in simulating tropical cyclones (TCs), their historical variability, and future change in a changing climate (Zhao et al. 2009). HiRAM is one of the GFDL models participating in CMIP5. HiRAM helped motivate the formation of the US CLIVAR Hurricane Working Group for a multi-institutional investigation of hurricane-climate connections using high-resolution GCMs. HiRAM also helped motivate the development of GFDL’s subseasonal to seasonal prediction system for tropical cyclones, MJO, and other extreme weather. I have published 5 lead-author papers using HiRAM and many co-authored papers, which are well-cited in the literature. In particular, the HiRAM documentation paper (Zhao et al. 2009) has been cited 435 (613) times based on Web of Science (Google Scholar). HiRAM has been used worldwide and has impacted many later works on TC-climate connections, TC seasonal predictions, global modeling of TC activities, TC intraseasonal variability. HiRAM simulation work contributed significantly to the GFDL Group Gold Medal awarded by the Department of Commerce in 2011 “for sustained high-quality research, scientific assessment and leadership resulting in an improved understanding of the impact of anthropogenic climate change on past and future hurricane activity”.

Development of GFDL’s latest generation global atmospheric model AM4, and coupled physical climate model CM4

I have co-led (2013-2015) and led (2015-2018) the GFDL Model Development Team (MDT) Atmospheric Working Group (AWG) for developing AM4 and co-led (2013-2019) the MDT Coupled Working Group (CWG) for developing CM4. This includes developing strategic plans, organizing meetings, analyzing and discussing model results, suggesting and creating new configurations and versions of AM4, developing and integrating new moist physics parameterizations, and diagnosing and addressing critical issues that surfaced during the AM4 development. My efforts on CM4 focused on reducing CM4’s biases in SSTs, ENSO, double ITCZ, as well as the global SST response to historical and present-day radiative forcing. One of the central goals was to reduce AM4/CM4 biases in simulations of climate through improved atmospheric moist physics. AM4 has been documented in Zhao et al. (2018a,b) and CM4 has been documented in Held et al. (2019). AM4 is a foundation for all new generation GFDL models including CM4, the Earth System Model (ESM4), and the latest GFDL subseasonal-seasonal to decadal prediction system (SPEAR). AM4, CM4, and ESM4 have participated in CMIP6. SPEAR has been running real-time for short-time climate prediction, and it has replaced the earlier GFDL prediction systems and contributed to the North American Multi-Model Ensemble.

Development of convective parameterization scheme and improvement for TC and MJO predictions

As a core developer of GFDL AM3, I implemented, further developed, and optimized the University of Washington Shallow Cumulus Scheme (UWShCu) and unified the plume model between UWShCu and Donner’s deep scheme to improve the model’s consistency and efficiency. My efforts led to a substantial improvement in AM3 simulation of climate. AM3 was awarded a Group Gold Medal by the DOC in 2012. During my development of HiRAM, I have further adapted UWShCu for representing both shallow and deep convection [See Zhao et al. (2009) Appendix for my modifications to UWShCu scheme as well as a simple statistical cloud scheme].  During my development of AM4/CM4, I  have further developed the convection scheme by including an additional deep plume to represent deep convection (Zhao et al. 2016, Zhao et al. 2018b). The new double plume convection (DPC) scheme emphasizes the importance of a non-intrusive convection parameterization, which allows a smoother transition between parameterized convection and explicit (large-scale) clouds and is responsible for many of the recent GFDL models’ improvements in simulating tropical transients such as MJO, tropical cyclones, and mesoscale convective systems. It also improves model simulations of mean precipitation, clouds, and cloud radiative effects. The DPC scheme has been used in not only the latest GFDL climate and earth system models (CM4, ESM4) but also GFDL’s latest prediction systems (SPEAR). When running in forecast mode, the DPC scheme has also been shown to substantially improve the models’ retrospective forecasts of MJO and TC genesis  (e.g., Xiang et al. 2021, Xiang et al. 2015a, Xiang et al. 2015b) at intra-seasonal scale. The DPC scheme has recently been adopted in a version of NCAR’s CAM5 model (Chu et al. 2021). 

Studies of clouds, cloud feedbacks, and climate sensitivities and co-leading GFDL Cloud Climate Initiative (CCI)

Since 2013, I have been co-leading the GFDL CCI. During this period, I have led three lead-author papers and many co-authored papers.  Zhao (2014) identified key physical processes (cumulus mixing and precipitation microphysics) and provided key diagnostic quantities (precipitation efficiency or cloud detrainment efficiency) in GCMs to understand the effects of convection on clouds and cloud feedbacks. This paper has motivated many later studies, including a chapter entitled “Precipitation efficiency and climate sensitivity” in the AGU Clouds and Climate Monograph Series (Lutsko et al. 2021). Zhao et al. (2016) further used a version of AM4 with changes only in the treatment of convective microphysics to demonstrate that convective precipitation microphysics (one of the most uncertain processes in GCM parameterizations) alone can profoundly affect cloud feedbacks and climate sensitivity, and its impact can be understood through precipitation efficiency. This paper received the 2018 NOAA OAR Outstanding Scientific Paper Award, and it helped motivate the NOAA CPO MAPP Climate Sensitivity Task Force, for which I am one of the co-leads. Most recently, Zhao (2021) has led an investigation of the equilibrium climate sensitivity (ECS) in GFDL’s latest climate models, CM4 and SPEAR. Using a series of coupled and uncoupled simulations, Zhao identified and quantified three major processes that have led to an increase in CM4’s ECS compared to earlier-generation GFDL models. These processes include changes in vegetation, sea-ice concentrations, and SST warming patterns. This paper demonstrated the limitations of the traditional Cess approach (i.e., uniform SST warming) in studies of cloud feedbacks and climate sensitivity and provided a  modified framework in understanding cloud feedbacks and climate sensitivity using atmosphere-only models. 

Studies of other high-impact weather [e.g., atmospheric rivers (AR), mesoscale convective systems (MCS); extreme cold weather; and storm-related extreme sea level] and their response to global warming

For the past few years, I have expanded my TC-climate studies to include other high-impact weather. For example, Zhao (2020) focused on atmospheric rivers (ARs), their variability, and change in warmer climates. It demonstrated the superior quality of the GFDL high-resolution AM4 simulations of present-day AR statistics and variability. Most previous studies of AR responses to global warming used a fixed threshold of integrated vapor transport (IVT) for detecting ARs and thus produced a large increase in the frequency of AR conditions in warmer climates. However, Zhao (2020) argued it is necessary to use an IVT threshold that accounts for the impact of global warming-induced moisture increases on the IVT threshold, and as a result, Zhao (2020) produced much less increase in AR frequency, but a much larger increase in AR intensity with warming. This paper was highlighted in the January 2021 issue of BAMS in the Papers of Note section. Zhao (2021) used satellite observations, reanalysis data, and high-resolution AM4 to quantify for the first time the collective role of AR, TS, and MCS in producing global and regional mean and extreme precipitation. The study not only demonstrates the model’s capability in simulating storm-associated precipitation and extreme precipitation but also reveals the changing character of storm-associated extreme precipitation in a warmer climate. This work has important implications for future flash flood-driven disasters and water resource management. In addition to the above two single-author papers, I have co-authored several other papers on weather-climate connections including an investigation of high-resolution AM4-simulated MCS climatology, variability, and response to global warming (Dong et al. 2020), a study of the effects of ocean circulation on US extreme cold weather (Yin and Zhao 2021), and a study of storm-related extreme sea level along the US Atlantic Coast (Yin et al. 2020).

Studies of tropical convection, clouds, and climate by developing and/or utilizing a model hierarchy

I have developed/utilized many idealized models to study convection, clouds, and climate. They include large-eddy-simulation (LES) models (e.g., Zhao and Austin 2005a,b), single-column models (SCMs, e.g., Zhao and Austin 2003), doubly periodic dynamical radiative-convective equilibrium (DRCE) models using GCM physics with (Held and Zhao 2008) and without ambient rotation (Held et al. 2007), aquaplanet models (APM, e.g. Kang et al. 2008, Medeiros et al. 2008, Merlis et al. 2013, 2016), uncoupled global atmospheric models (AGCMs, e.g., Zhao 2014, Zhao et al. 2016), and fully coupled global climate models (CGCMs, e.g., Zhao 2021) to explore convection, clouds, and their relationship to the large-scale environment. My work and the idealized models I developed have motivated and helped many graduate students and postdocs at GFDL and Princeton University to use the idealized modeling frameworks for their research and development.

Selected Publications Based on Areas of Research