Improved Understanding of air-sea interaction processes and biases in the Tropical Western Pacific using observation sensitivity experiments and global forecast models

PIs: Dr. Aneesh Subramanian (CU Boulder), Dr. Kris Karnauskas (CU Boulder), Dr. Charlotte DeMott (CSU Fort Collins), Dr. Matthew Mazloff (SIO, UCSD)
Post-Doc: Dr. Ho-Hsuan Wei (CU Boulder)

Description

The team studies the physical mechanisms governing air-sea interactions in the tropical west Pacific at the eastern edge of the warm pool by isolating coupled feedback processes through analyses of long climate model runs, reanalyses products and short-term coupled and uncoupled forecasts. Climate model forecasts of the Madden–Julian Oscillation (MJO) and El Niño–Southern Oscillation (ENSO) experience a systematic climate drift resulting in biases of the modeled tropical western Pacific climatology. Global models tend to have an excess rainfall in the warm pool region and a deficiency in rainfall at the eastern edge of the warm pool. We use the Community Earth System Model (CESM) global runs as well as high-resolution Massachusetts Institute of Technology general circulation model (MITgcm) regional uncoupled simulations to study ocean heat, salinity and momentum redistributions in the region. The oceanic barrier layer at the eastern edge of the warm pool can play a dynamical role in modulating air-sea interaction over the region influencing MJO and ENSO evolution over the Tropical Pacific region. The schematic below shows the local and remote impacts of temperature and salinity stratification in the West Pacific on ENSO development.

This research will identify the physical processes that lead to the mean and intraseasonal variability biases in the tropical west Pacific and, more specifically, the key biases controlling MJO and ENSO background states. This understanding prioritizes process studies and observational field campaigns in the region as a part of TPOS 2020. The work will also inform parameterization improvements for use in CESM for real-time forecasts of the climate-scale processes in the tropical Pacific.

This project was funded by the NOAA’s Climate Variability and Predictability Program (CVP) as one of its eight pre-field campaign modeling projects in support NOAA’s contribution to TPOS 2020.

Accomplishments

We have performed CESM 2 simulations to characterize the model mean state bias in several variables in the Tropical Pacific region. We have completed two sets of 25 years of simulation with CESM 2, one with pre-industrial forcing and another with 20th century radiative forcing. We have saved high frequency (daily) ocean output from these simulations to diagnose tropical:

• mixed layer depth variability
• barrier layer formation
• warm pool extension

The goal of these diagnostics is to document model climatology and inherent MJO skill in advance of forecast experiments.

We have also obtained ocean reanalysis data from ORA-S5 (ECMWF) reanalysis product for comparing the ocean model (POP) output from CESM2 simulations. The comparisons of the surface variables (temperature, salinity and precipitation) are shown in Figure 1. Here we see large biases in the Tropical West Pacific warm pool region. The annual mean CESM2 biases in western Pacific show warm and fresh surface biases. The warm pool expands eastward and the barrier layer thickness (BLT) is too shallow (shown in Figure 2a).”

We have examined sources of SST drift in coupled forecast models that have provided data to the international S2S Database. Climatological SST drift (i.e., the November-April averaged change in mean state SST as a function of lead time, dSST/dt) was decomposed into one component correlated with the net surface heat flux (Qnet) and a second component associated with ocean processes. The Qnet-driven SST drift was estimated by regressing the climatological dSST/dt onto climatological Qnet; the ocean component of drift is the dSST/dt that is not correlated with Qnet, i.e., the residual. Maps of Qnet and ocean dynamics contributions to 30-day SST drift in S2S models shown in Fig. 3 reveal wide discrepancies in regional sources of SST drift among models. Lead-dependent time series of each term (not shown) sometimes reveals large initial contributions of Qnet or ocean dynamics terms that reflect initialization shock, whereas monotonic tendencies of each term over the 30-day forecast period reflect drift toward models’ preferred mean states.

Lessons Learned

The subsurface ocean features are important for air-sea interaction and have been shown to play important roles in the development of the El Nino events. The barrier layer and isothermal layer distribution across the Tropical Pacific and their evolution can play an important role in modulating air-sea interaction over the region and hence influence ENSO evolution. Hence, observing the variability of the barrier layer over the eastern edge of the warm pool region, mixing processes in the region and the variability and drivers of the barrier layer would be very informative to guide model development in improving these process representations.

Publications

Wei, H-H., A. C. Subramanian, K. Karnauskas, C. A. DeMott; M. R. Mazloff; M. A. Balmaseda, (2020): Tropical Pacific Air-sea Interaction Processes and Biases in CESM2 and their Relation to El Nino Development. J. Clim., sub judice.

Jiang, X., A. Adames-Corraliza, A., D. Kim, E. Maloney, H. Lin, H. Kim, C. Zhang, C. DeMott, and N. Klingaman, 2020: Fifty Years of Research on the Madden-Julian Oscillation: Recent Progress, Challenges, and Perspectives, Accepted to JGR Atmospheres.

Fredriksen, H-B., J. Berner, A. C. Subramanian, A. Capotondi (2020): How Does El Niño Southern Oscillation Change Under Global Warming – A First Look at CMIP6, GRL, 47, e2020GL090640. https://doi.org/10.1029/2020GL090640.

Karnauskas, K. B. (2021) A simple coupled model of the wind-evaporation-SST feedback with a role for stability. J. Climate, submitted.

Verdy, A., M. Mazloff, B. D. Cornuelle, A. C. Subramanian (2021) ENSO influence on heat and freshwater budgets in the tropical Pacific Ocean state estimate, to be submitted.

Data

Data can be downloaded from the Tropical Pacific Ocean State Estimation website

Figure 2: The longitudinal distribution of the (a) BLT, (b) ILD (dashed) and MLD (solid) for ORAS5 (black) and CESM2 (red).

Figure 3: SST “balance factor” (following Halkides et al. 2015; doi:10.1002/2014JC010139 ) for sources of climatological SST drift in S2S database coupled forecast models.
Warm colors correspond to Qnet-driven drift, while cool colors correspond to ocean dynamics-driven drift.