Convective self-aggregation: growing storms in a virtual laboratory

By: Chris Holloway

Figure 1: An example of convective self-aggregation from an RCE simulation using the Met Office Unified Model at 4km grid length with 300 K SST.  Time mean precipitation in mm/day for (a) Day 2 (still scattered), and (b) Day 40 (aggregated).  Note that the lateral boundaries are bi-periodic, so the cluster in (b) is a single organised region.  Adapted from Holloway and Woolnough (2016).

Convective self-aggregation is the clumping together of isolated convective cells (rainstorms) into organised regions in idealised computer simulations.  This storm clustering may not seem all that unusual, but it is surprising because self-aggregation occurs in simulations of “radiative-convective equilibrium” (RCE) in which boundary conditions are homogeneous (sea surface temperature [SST] is constant in space and time), there is no imposed mean wind, and planetary rotation is set to zero (e.g., Figure 1).  In other words, there is no external cause of the clustering of convection in self-aggregation (hence the “self” prefix).  Instead, internal feedbacks such as cloud-radiation interactions and surface-flux feedbacks are key (Wing et al.2017 and references therein).

Figure 2: Satellite estimates of average fractional cover vs total Cold Cloud Area for a given domain-mean precipitation rate (R) range and for ranges of the “SCAI” aggregation index between 0.00 and 0.35 (red, aggregated), between 0.35 and 0.70 (black, intermediate), and between 0.70 and 1.50 (blue, disaggregated); t stands for optical thickness. Shaded regions indicate the 90% confidence interval. (a)–(c) thick anvil, (d)–(f) optically thin anvil.  Adapted from Stein et al. (2017).

While self-aggregation is intellectually interesting, many scientists are sceptical of the relevance of this phenomenon for real weather and climate.  After all, the real world has plenty of inhomogeneity in surface temperature as well as rotation and vertical wind shear.  However, organised tropical convection in real-world observations shows many similarities to self-aggregated convection in idealised simulations: for more aggregated conditions the mean state has lower relative humidity, outgoing longwave radiation (OLR) is larger, and anvil cloud amount is reduced (Holloway et al. 2017 and references therein).  For instance, work at the University of Reading using satellite observations has shown that optically thin anvil cloud cover decreases as convection becomes more aggregated, which could have implications for climate (Figure 2).    More realistic convective-scale simulations of organised tropical convection (with observed SSTs, rotational effects, and wind shear effects) also provide evidence that cloud-radiation feedbacks act to maintain organisation and reduce the mean relative humidity (Holloway 2017).   Other real-world forms of organised tropical convection, including the Madden-Julian Oscillation (MJO), tropical cyclones, and the Intertropical Convergence Zone (ITCZ) all show cloud-radiative feedbacks and moisture-convection feedbacks that resemble processes important for convective self-aggregation in idealised computer simulations.

The potential impact of convective aggregation on climate is an area of active research and debate.  Some idealised computer experiments show stronger self-aggregation with warmer SSTs sea surfaces, but others do not (Wing 2019).  If aggregation were to increase with increasing SST, this would likely be a slightly negative feedback for global warming, meaning it would allow for slightly less warming for a given increase in carbon dioxide concentrations, but this is also an active area of research and debate.  Aggregation tends to be weaker and more variable in simulations that include coupled ocean models (e.g. Hohenegger and Stevens 2016, Coppin and Bony 2017), so this is another area that needs more extensive research. 

Studying convective aggregation enables the scientific community to generate and test hypotheses and isolate mechanisms about fundamental processes that are potentially important for convective organisation, but which can be difficult to disentangle in more realistic settings.  Even if studies eventually demonstrate how self-aggregation is not an adequate framework for some climate problems, this will also be a form of important progress.  The Radiative-Convective Equilibrium Model Intercomparison Project (RCEMIP) is bringing scientific institutions together to compare self-aggregation at different model resolutions, domains and SSTs in order to facilitate further research into this exciting topic.  At Reading, Met Office Unified Model convection-permitting simulations have been performed and submitted to RCEMIP in association with the joint NERC-Met Office ParaCon project which seeks to greatly improve the representation of convection in weather and climate models.   RCEMIP and other research efforts will increasingly apply new concepts emerging from idealised simulations to the complex interactions between convection, moisture, clouds, radiation, surface fluxes, circulations and climate.


Coppin, D., and S. Bony, 2017: Internal variability in a coupled general circulation model in radiative‐convective equilibrium, Geophysical Research Letters, 44, 10, 5142-5149,

Hohenegger, C., and Stevens, B., 2016: Coupled radiative convective equilibrium simulations with explicit and parameterized convection. J. Adv. Model. Earth Syst., 8, 1468–1482, doi:10.1002/2016MS000666.

Holloway, C. E., 2017: Convective aggregation in realistic convective- scale simulations, J. Adv. Model. Earth Syst., 9, 1450–1472, doi:10.1002/ 2017MS000980.

Holloway, C. E., A. A. Wing, S. Bony, C. Muller, H. Masunaga, T. S. L’Ecuyer, D. D. Turner, and P. Zuidema, 2017: Observing convective aggregation.  Surveys in Geophysics, 38: 1199. doi:10.1007/s10712-017-9419-1. 

Holloway, C. E., and S. J. Woolnough, 2016: The sensitivity of convective aggregation to diabatic processes in idealized radiative-convective equilibrium simulations.  J. Adv. Model. Earth Syst., 8, 166–195, doi:10.1002/2015MS000511.

Stein, T. H. M., C. E. Holloway, I. Tobin, and S. Bony, 2017: Observed relationships between cloud vertical structure and convective aggregation over tropical ocean.  J. Climate, 30, 2187–2207. 

Wing, A. A., 2019: Self-Aggregation of Deep Convection and its Implications for Climate. Curr. Clim. Change Rep., 5: 1.

Wing, A. A., K. Emanuel, C. E. Holloway, and C. Muller, 2017: Convective self-aggregation in numerical simulations: A review.  Surveys in Geophysics, 38: 1173.



This entry was posted in Climate, Climate modelling, Numerical modelling, Tropical convection. Bookmark the permalink.