Periodic Reporting for period 4 - ECLAIR (Emulation of subgrid-scale aerosol-cloud interactions in climate models: towards a realistic representation of aerosol indirect effect)
Okres sprawozdawczy: 2020-03-01 do 2020-08-31
ECLAIR aimed to develop an innovative interdisciplinary model framework to refine the estimate of aerosol indirect effect (i.e. influence of atmospheric aerosol particles on cloud properties), which remains the single largest uncertainty in the current drivers of climate change. A major reason for this uncertainty is that current climate models are unable to resolve the spatial scales for aerosol-cloud interactions.
We proposed to resolve this scale problem by using statistical emulation to build computationally fast surrogate models (i.e. emulators) that can reproduce the effective output of a detailed high-resolution cloud-scale model. By incorporating these emulators into a state-of-the-science climate model, the aim was to, for the first time, achieve the accuracy of a limited-area high-resolution model on a global scale with negligible computational cost.
The main scientific outcome of the project was anticipated to be a highly refined and physically sound estimate of the aerosol indirect effect that enables more accurate projections of future climate change, and thus has high societal relevance. In addition, the work aimed at quantifying largest uncertainties in and improving the process-level understanding of aerosol-cloud interactions.
The key project objectives were achieved. The main breakthrough of the project was the development of the pioneering emulator-based approach in climate modelling. We expect this approach to open up completely new research opportunities also in other fields that deal with heterogeneous spatial scales. In addition, the project developed arguably the most detailed cloud-scale model in the world in terms of aerosol-cloud-precipitation interactions. With the help of this model, several aspects of process-level understanding have already been improved.
We proposed to resolve this scale problem by using statistical emulation to build computationally fast surrogate models (i.e. emulators) that can reproduce the effective output of a detailed high-resolution cloud-scale model. By incorporating these emulators into a state-of-the-science climate model, the aim was to, for the first time, achieve the accuracy of a limited-area high-resolution model on a global scale with negligible computational cost.
The main scientific outcome of the project was anticipated to be a highly refined and physically sound estimate of the aerosol indirect effect that enables more accurate projections of future climate change, and thus has high societal relevance. In addition, the work aimed at quantifying largest uncertainties in and improving the process-level understanding of aerosol-cloud interactions.
The key project objectives were achieved. The main breakthrough of the project was the development of the pioneering emulator-based approach in climate modelling. We expect this approach to open up completely new research opportunities also in other fields that deal with heterogeneous spatial scales. In addition, the project developed arguably the most detailed cloud-scale model in the world in terms of aerosol-cloud-precipitation interactions. With the help of this model, several aspects of process-level understanding have already been improved.
The cloud-scale model UCLALES-SALSA was significantly improved during the project, e.g. by implementing ice-phase microphysics into the model and refining the treatment of aerosol particle growth. In addition, the model was comprehensively tested against a wide range of in-situ and satellite observations. This development and evaluation work has greatly improved the process understanding of several aspects of aerosol-cloud interactions and has made UCLALES-SALSA arguably the most detailed cloud-scale model in the world. This work benefits scientists interpreting experimental cloud data and predicting cloud responses to changes in aerosol emissions and environmental conditions.
For the first time, an emulator-based approach for prediction of aerosol indirect effects was developed and implemented in a climate model. The approach provides a clearly improved accuracy over conventional parameterizations in many regions of the world. This is an important advancement for the climate modelling community and the main breakthrough in the project. Simulations with the new model framework highlight the importance of accurate representation of aerosol-cloud interactions both to regional and global climate predictions, which has direct relevance to climate change impacts and adaptation. It is anticipated that the developed approach will open up new research opportunities also in other fields of science with highly heterogeneous spatial scales.
Project findings have been presented in number of scientific meetings, workshops and lectures and their potential use in other applications has been discussed with several researchers in the atmospheric sciences field. After the key publications detailing the novel model framework have been completed (manuscripts in preparation), we intend to widen our dissemination activities also outside our own field to further spread knowledge of the developed methodology that is potentially applicable to a wide range of scientific computing problems. The developed cloud-scale model is openly accessible for scientific use in GitHub under MIT license and the pioneering climate model framework will be shared openly with collaborators.
For the first time, an emulator-based approach for prediction of aerosol indirect effects was developed and implemented in a climate model. The approach provides a clearly improved accuracy over conventional parameterizations in many regions of the world. This is an important advancement for the climate modelling community and the main breakthrough in the project. Simulations with the new model framework highlight the importance of accurate representation of aerosol-cloud interactions both to regional and global climate predictions, which has direct relevance to climate change impacts and adaptation. It is anticipated that the developed approach will open up new research opportunities also in other fields of science with highly heterogeneous spatial scales.
Project findings have been presented in number of scientific meetings, workshops and lectures and their potential use in other applications has been discussed with several researchers in the atmospheric sciences field. After the key publications detailing the novel model framework have been completed (manuscripts in preparation), we intend to widen our dissemination activities also outside our own field to further spread knowledge of the developed methodology that is potentially applicable to a wide range of scientific computing problems. The developed cloud-scale model is openly accessible for scientific use in GitHub under MIT license and the pioneering climate model framework will be shared openly with collaborators.
The project developed a pioneering emulator-based approach for prediction of aerosol-cloud effects, which provides improved accuracy over conventional parameterizations in climate models. This is an important advancement for the climate modelling community. The unique new approach can be extended to a wide range of scientific modelling applications in various fields that deal with varying spatial scales. The project also set a new bar for detailing aerosol-cloud-precipitation interactions in cloud-scale models, with the model version developed in this project now arguably the most detailed model in the world.