Although a simple form of Machine learning (ML), namely artificial neural networks (ANN), has been used extensively to forecast convective hazards since the mid-1990s, ANN has been often criticized by forecasters and end-users as being a “black box” because of the perceived inability to understand how ML makes its predictions. Recently there is a strong interest to explore the use of interpretable ML methods, often used in Data Science, for climate predictions and projections.
This Research Topic will explore the questions: Are ML techniques adequate for climate predictions and projections? Is there any value-added to the existing methods? What are the opportunities and challenges for using ML in climate prediction and projection studies?
The goal of this topic is to document the application of interpretable ML as a value-added over popular methods for climate prediction and projections. For the climate prediction, the focus will be on the Subseasonal-to-Seasonal (S2S) forecasting (between two weeks and a season ahead) which is a rapidly developing area of forecast development, with the potential to provide valuable information for the development of climate services. On the climate projection side, ML-based downscaling on the new CMIP6 experiments will be the focus.
Recently, ML has been proposed for representing physical processes and parameterization schemes in a dynamical modeling system. Research related to ML techniques in model parameterization schemes in climate modeling is also a core area for this topic.
This topic will also encourage to submit of research papers focusing on developing user-friendly tools for ML to use in climate prediction and projection.
Submissions are welcomed on the following themes and related areas:
• ML-based post-processing (bias correction, multi-model ensemble) of the GCM’s output to improve the Subseasonal-to-Seasonal (S2S) forecast for extreme events (drought/flood, heat/cold wave).
• ML-based statistical downscaling of CMIP6 based climate change projection.
• ML-based statistical modeling (time series, teleconnection, etc.) for S2S forecast and climate change projection.
• ML techniques in model parameterization schemes in climate modelling system.
• ML techniques to calibrate or develop model parameterization schemes in climate modelling systems.
• Development of software or packages (Python, R or MATLAB) for climate predictions and projections studies.
Although a simple form of Machine learning (ML), namely artificial neural networks (ANN), has been used extensively to forecast convective hazards since the mid-1990s, ANN has been often criticized by forecasters and end-users as being a “black box” because of the perceived inability to understand how ML makes its predictions. Recently there is a strong interest to explore the use of interpretable ML methods, often used in Data Science, for climate predictions and projections.
This Research Topic will explore the questions: Are ML techniques adequate for climate predictions and projections? Is there any value-added to the existing methods? What are the opportunities and challenges for using ML in climate prediction and projection studies?
The goal of this topic is to document the application of interpretable ML as a value-added over popular methods for climate prediction and projections. For the climate prediction, the focus will be on the Subseasonal-to-Seasonal (S2S) forecasting (between two weeks and a season ahead) which is a rapidly developing area of forecast development, with the potential to provide valuable information for the development of climate services. On the climate projection side, ML-based downscaling on the new CMIP6 experiments will be the focus.
Recently, ML has been proposed for representing physical processes and parameterization schemes in a dynamical modeling system. Research related to ML techniques in model parameterization schemes in climate modeling is also a core area for this topic.
This topic will also encourage to submit of research papers focusing on developing user-friendly tools for ML to use in climate prediction and projection.
Submissions are welcomed on the following themes and related areas:
• ML-based post-processing (bias correction, multi-model ensemble) of the GCM’s output to improve the Subseasonal-to-Seasonal (S2S) forecast for extreme events (drought/flood, heat/cold wave).
• ML-based statistical downscaling of CMIP6 based climate change projection.
• ML-based statistical modeling (time series, teleconnection, etc.) for S2S forecast and climate change projection.
• ML techniques in model parameterization schemes in climate modelling system.
• ML techniques to calibrate or develop model parameterization schemes in climate modelling systems.
• Development of software or packages (Python, R or MATLAB) for climate predictions and projections studies.
