Periodic Reporting for period 4 - INTERACTION (Cloud-cloud interaction in convective precipitation)
Okres sprawozdawczy: 2023-01-01 do 2024-04-30
Thunderstorm clouds are observed to cluster in space, which can lead to tropical cyclones and extreme precipitation events both in the tropics and in mid-latitudes. This clustering is observed from satellite and ground based instruments, and can be simulated, but the fundamental mechanisms for why it occurs, are not well-understood.
Society is most directly affected by extreme precipitation events from clustered thunderstorms, as these can lead to flash floods - floods occurring as a result of heavy precipitation (downpours) over a relatively short period (few hours). Flash floods can be devastating to human lives and infrastructure, and appear to occur more frequently over recent years. Understanding the mechanisms that lead to flash floods will allow for better predictions of flood risk, today and in the future climate. Furthermore, cloud clusters can influence, how Earth's atmosphere interacts with radiation, such as sunlight and thermal radiation. A change in clustering under increasing surface temperatures could feed back on the heating through changes in the radiation that reaches and leaves the surface - thus again modifying the climate change signal.
The project has the overall objective to develop a clear understanding of how thunderstorms cluster. The working hypothesis is, that convective clouds interact through so-called "cold pools," air masses cooled by rainfall beneath thunderstorm clouds. When cold pools collide as they spread along the surface, they can excite new thunderstorms. This mechanism encodes a spatial interaction effect, which we describe theoretically and by conceptual modeling. This will lead to novel convective parameterizations in climate models, which in turn can yield more reliable future climate projections.
Society is most directly affected by extreme precipitation events from clustered thunderstorms, as these can lead to flash floods - floods occurring as a result of heavy precipitation (downpours) over a relatively short period (few hours). Flash floods can be devastating to human lives and infrastructure, and appear to occur more frequently over recent years. Understanding the mechanisms that lead to flash floods will allow for better predictions of flood risk, today and in the future climate. Furthermore, cloud clusters can influence, how Earth's atmosphere interacts with radiation, such as sunlight and thermal radiation. A change in clustering under increasing surface temperatures could feed back on the heating through changes in the radiation that reaches and leaves the surface - thus again modifying the climate change signal.
The project has the overall objective to develop a clear understanding of how thunderstorms cluster. The working hypothesis is, that convective clouds interact through so-called "cold pools," air masses cooled by rainfall beneath thunderstorm clouds. When cold pools collide as they spread along the surface, they can excite new thunderstorms. This mechanism encodes a spatial interaction effect, which we describe theoretically and by conceptual modeling. This will lead to novel convective parameterizations in climate models, which in turn can yield more reliable future climate projections.
INTERACTION focused on how thunderstorm clouds influence each other through their hydrodynamics and thermodynamics. Hydrodynamically, a convective cell is characterized by initially rapid updrafts, subsequent rain formation and finally a deep convective cold pool (CP). CPs are caused by rain evaporation in the subcloud layer and often give rise to density currents spreading along the surface. The momentum transport can cause subsequent lifting and convection in the vicinity. Another mechanism, thermodynamically driven, is that of buoyancy transport through CPs.
Work within INTERACTION addressed the different effects mediated by CPs:
- the small-scale interaction mechanisms between CPs and the resulting self-organized structures.
- isolated interaction effects, studying the resolution dependence of CPs and their dynamical properties as well as the collision mechanism.
- tracking of CPs in models and observations, e.g. from satellite, using particle-based dynamical methods and machine-learning based statistical methods to distinguish CP from non-CP regions.
- emergent self-organization: INTERACTION employed a range of approaches, from conceptual models, e.g. a "circle model" where cold pools are effectively modeled as spreading circles that are able to collide, to more "coarse-grained" models where the collision field between cold pools is described to cloud-resolving simulations of CPs organizing under the diurnal cycle and over ocean. Also wind shear was imposed to mimic a realistic large-scale forcing for the tropical or subtropical atmosphere. Important findings included "diurnal self-aggregation", a novel insight into so-called "convective self-aggregation".
In INTERACTION, several new routes to convective self-aggregation were identified: (a) that interacting CPs together with a mesoscale energetic constraint can already give rise to strongly aggregated atmospheres where only a part of the domain is covered by convection whereas other parts are cloud free; (b) that convective self-aggregation can be induced under conditions of relatively high horizontal resolution (~1km) when combined with a surface temperature diurnal cycle, giving rise to strong CP effects and organization into mesoscale convective systems. This self-organization does not take place when the diurnal cycle is removed. In addition, once organized, strong clustering persists when the diurnal cycle is switched off - mimicking a transition from land to sea.
Since 2023, INTERACTION has contributed to a dedicated field campaign "High-resolution weather observations east of Dakar (DakE)" in Senegal, a region well-known for its convective self-organization effects: mesoscale convective systems are common during the rainy season and drive extreme rainfall - leading to flood risk in metropolitan areas, e.g. Dakar. The team has been implementing an observational network east of Dakar with more than 15 automatic weather stations as well as ~100 additional measurement devices including soil, flood and humidity sensors.
Several conferences were organized in INTERACTION: Niels Bohr Institute, U Copenhagen (May 5-7, 2001), Utrecht (2002) and ICTP Trieste (2003 & 2004). The "Workshop of Convective Organization" is now a yearly recurring event drawing ~100 scientists from around the world. In 2026 the event is taking place in Brazil.
Work within INTERACTION addressed the different effects mediated by CPs:
- the small-scale interaction mechanisms between CPs and the resulting self-organized structures.
- isolated interaction effects, studying the resolution dependence of CPs and their dynamical properties as well as the collision mechanism.
- tracking of CPs in models and observations, e.g. from satellite, using particle-based dynamical methods and machine-learning based statistical methods to distinguish CP from non-CP regions.
- emergent self-organization: INTERACTION employed a range of approaches, from conceptual models, e.g. a "circle model" where cold pools are effectively modeled as spreading circles that are able to collide, to more "coarse-grained" models where the collision field between cold pools is described to cloud-resolving simulations of CPs organizing under the diurnal cycle and over ocean. Also wind shear was imposed to mimic a realistic large-scale forcing for the tropical or subtropical atmosphere. Important findings included "diurnal self-aggregation", a novel insight into so-called "convective self-aggregation".
In INTERACTION, several new routes to convective self-aggregation were identified: (a) that interacting CPs together with a mesoscale energetic constraint can already give rise to strongly aggregated atmospheres where only a part of the domain is covered by convection whereas other parts are cloud free; (b) that convective self-aggregation can be induced under conditions of relatively high horizontal resolution (~1km) when combined with a surface temperature diurnal cycle, giving rise to strong CP effects and organization into mesoscale convective systems. This self-organization does not take place when the diurnal cycle is removed. In addition, once organized, strong clustering persists when the diurnal cycle is switched off - mimicking a transition from land to sea.
Since 2023, INTERACTION has contributed to a dedicated field campaign "High-resolution weather observations east of Dakar (DakE)" in Senegal, a region well-known for its convective self-organization effects: mesoscale convective systems are common during the rainy season and drive extreme rainfall - leading to flood risk in metropolitan areas, e.g. Dakar. The team has been implementing an observational network east of Dakar with more than 15 automatic weather stations as well as ~100 additional measurement devices including soil, flood and humidity sensors.
Several conferences were organized in INTERACTION: Niels Bohr Institute, U Copenhagen (May 5-7, 2001), Utrecht (2002) and ICTP Trieste (2003 & 2004). The "Workshop of Convective Organization" is now a yearly recurring event drawing ~100 scientists from around the world. In 2026 the event is taking place in Brazil.
INTERACTION has led to a novel understanding of convective self-organization and its modeling.
E.g. conceptual works on convective self-organization used "circle models" applied to the diurnal cycle over land as well as the sea, with theoretical insight into convective self-organization by interactions. We constructed a reaction‐diffusion model, that mimics a tipping process found in the formation of MCS under the diurnal in cloud-resolving models. Results suggest strong cold pool interactions brought about by the diurnal cycle. Diurnal cycle effects on mesoscale convective systems (MCS) and the relation to classical self-aggregation were analyzed. The diurnal cycle has profound effects on mesoscale self-organization and sustainably structures the moisture field - inducing a moisture-radiation-circulation feedback and hysteresis.
By studying mesh resolution in simulations, the influence of resolution on cold pools (CPs) was examined. Core findings include that lobe-and-cleft instabilities at the gust front require 100m resolution whereas collision effects can reasonably be simulated at a 1km resolution. This work serves as a benchmark for parameterizations.
A comprehensive, and novel, characterization of MCS over Europe tracked MCS and analyzed them throughout Europe and addressed relevance for extreme precipitation. A marked convective diurnal cycle is found, with daytime and nocturnal peaks. A related work revisited the question of the Clausius-Clapeyron rate and its potential exceedence for convective extremes.
A CP detection and tracking algorithm (CoolDeTA) was developed to identify CPs and track them throughout their lifecycle allowing for identification of active gust fronts and their “offspring” to reconstruct CP family trees. To make identification of CPs possible on a global scope, we develop a methodology to detect cold pools using data at (a) global availability and (b) high spatiotemporal resolution. For this, a convolutional neural networks is trained to segment CPs in high‐resolution cloud‐resolving data. We compare station-based identifications of CPs with corresponding satellite data and find good agreement. The study suggests that CPs might be identifiable from satellite data when conditioned properly, a feature we will exploit by combining with our DakE campaign.
E.g. conceptual works on convective self-organization used "circle models" applied to the diurnal cycle over land as well as the sea, with theoretical insight into convective self-organization by interactions. We constructed a reaction‐diffusion model, that mimics a tipping process found in the formation of MCS under the diurnal in cloud-resolving models. Results suggest strong cold pool interactions brought about by the diurnal cycle. Diurnal cycle effects on mesoscale convective systems (MCS) and the relation to classical self-aggregation were analyzed. The diurnal cycle has profound effects on mesoscale self-organization and sustainably structures the moisture field - inducing a moisture-radiation-circulation feedback and hysteresis.
By studying mesh resolution in simulations, the influence of resolution on cold pools (CPs) was examined. Core findings include that lobe-and-cleft instabilities at the gust front require 100m resolution whereas collision effects can reasonably be simulated at a 1km resolution. This work serves as a benchmark for parameterizations.
A comprehensive, and novel, characterization of MCS over Europe tracked MCS and analyzed them throughout Europe and addressed relevance for extreme precipitation. A marked convective diurnal cycle is found, with daytime and nocturnal peaks. A related work revisited the question of the Clausius-Clapeyron rate and its potential exceedence for convective extremes.
A CP detection and tracking algorithm (CoolDeTA) was developed to identify CPs and track them throughout their lifecycle allowing for identification of active gust fronts and their “offspring” to reconstruct CP family trees. To make identification of CPs possible on a global scope, we develop a methodology to detect cold pools using data at (a) global availability and (b) high spatiotemporal resolution. For this, a convolutional neural networks is trained to segment CPs in high‐resolution cloud‐resolving data. We compare station-based identifications of CPs with corresponding satellite data and find good agreement. The study suggests that CPs might be identifiable from satellite data when conditioned properly, a feature we will exploit by combining with our DakE campaign.