Search

Machine learning (ML) algorithms can be used in Earth system models (ESMs) to emulate sub-grid-scale processes. Due to the statistical nature of ML algorithms and the high complexity of ESMs, these hybrid ML ESMs require careful validation. Simulation stability needs to be monitored in fully coupled simulations, and the plausibility of results needs to be evaluated in suitable experiments. We present the coupling of SuperdropNet, a machine learning model for emulating warm-rain processes in cloud microphysics, with ICON (Icosahedral Nonhydrostatic) model v2.6.5. SuperdropNet is trained on computationally expensive droplet-based simulations and can serve as an inexpensive proxy within weather prediction models. SuperdropNet emulates the collision–coalescence of rain and cloud droplets in a warm-rain scenario and replaces the collision–coalescence process in the two-moment cloud microphysics scheme. We address the technical challenge of integrating SuperdropNet, developed in Python and PyTorch, into ICON, written in Fortran, by implementing three different coupling strategies: embedded Python via the C foreign function interface (CFFI), pipes, and coupling of program components via Yet Another Coupler (YAC). We validate the emulator in the warm-bubble scenario and find that SuperdropNet runs stably within the experiment. By comparing experiment outcomes of the two-moment bulk scheme with SuperdropNet, we find that the results are physically consistent and discuss differences that are observed in several diagnostic variables. In addition, we provide a quantitative and qualitative computational benchmark for three different coupling strategies – embedded Python, coupler YAC, and pipes – and find that embedded Python is a useful software tool for validating hybrid ML ESMs. [ABSTRACT FROM AUTHOR]

Machine learning (ML) algorithms can be used in Earth System models (ESMs) to emulate sub-grid-scale processes. Due to the statistical nature of ML algorithms and the high complexity of ESMs, these hybrid ML-ESMs require careful validation. Simulation stability needs to be monitored in fully coupled simulations, and the plausibility of results needs to be evaluated in suitable experiments. We present the coupling of SuperdropNet, a machine learning model for emulating warm rain processes in cloud microphysics, into ICON~2.6.5. SuperdropNet is trained on superdroplet simulations and predicts updates of the bulk moments for cloud and rain. It replaces the accretion, autoconversion, and self-collection of rain and cloud droplets in two-moment cloud microphysics. We address the technical challenge of integrating SuperdropNet, developed in Python and PyTorch, into ICON, written in Fortran, by implementing three different coupling strategies: embedded Python via the C Foreign Function Interface, pipes, and coupling of program components via YetAnotherCoupler (YAC). We validate the emulator in the warm bubble scenario and find that SuperdropNet runs stable within the experiment. In comparing experiment outcomes from the bulk moment scheme and SuperdropNet, we find that the results are physically consistent, and discuss differences that are observed for several diagnostic variables. In addition, we provide a quantitative and qualitative computational benchmark for three different coupling strategies—embedded Python, coupler YAC, and pipes—and find that embedded Python is a useful software tool for validating hybrid ML-ESMs. [ABSTRACT FROM AUTHOR]

The current handling of data in earth observation, modelling and prediction measures gives cause for critical consideration, since we all too often carelessly ignore data uncertainty. We think that Earth scientists are generally aware of the importance of linking data to quantitative uncertainty measures. But we also think that uncertainty quantification of Earth observation data too often fails at very early stages. We claim that data acquisition without uncertainty quantification is not sustainable and machine learning and computational modelling cannot unfold their potential when analysing complex natural systems like the Earth. Current approaches such as stochastic perturbation of parameters or initial conditions cannot quantify uncertainty or bias arising from the choice of model, limiting scientific progress. We need incentives stimulating the honest treatment of uncertainty starting during data acquisition, continuing through analysis methodology and prediction results. Computational modellers and machine learning experts have a critical role, since they enjoy high esteem from stakeholders and their methodologies and their results critically depend on data uncertainty. If both want to advance their uncertainty assessment of models and predictions of complex systems like the Earth, they have a common problem to solve. Together, computational modellers and machine learners could develop new strategies for bias identification and uncertainty quantification offering a more all‐embracing uncertainty quantification than any known methodology. But since it starts for computational modellers and machine learners with data and their uncertainty, the fundamental first step in such a development would be leveraging shareholder esteem to insistently advocate for reduction of ignorance when it comes to uncertainty quantification of data. [ABSTRACT FROM AUTHOR]

Recently, Machine learning (ML) has been widely utilized for laboratory earthquake (labquake) prediction using various types of data. This study pioneers in time to failure (TTF) prediction based on ML using acoustic emission (AE) records from three laboratory stick-slip experiments performed on Westerly granite samples with naturally fractured rough faults, more similar to the heterogeneous fault structures in the nature. 47 catalog-driven seismo-mechanical and statistical features are extracted introducing some new features based on focal mechanism. A regression voting ensemble of Long-Short Term Memory (LSTM) networks predicts TTF with a coefficient of determination (R 2) of 70% on the test dataset. Feature importance analysis revealed that AE rate, correlation integral, event proximity, and focal mechanism-based features are the most important features for TTF prediction. Results reveal that the network uses all information among the features for prediction, including general trends in high correlated features as well as fine details about local variations and fault evolution involved in low correlated features. Therefore, some highly correlated and physically meaningful features may be considered less important for TTF prediction due to their correlation with other important features. Our study provides a ground for applying catalog-driven to constrain TTF of complex heterogeneous rough faults, which is capable to be developed for real application. • Catalog-driven features are informative for preparatory phase characterization. • ML can predict TTF using physics-informed features in natural rough faults. • Feature importance analysis reveals most important features for TTF prediction. • AE rate, correlation integral, proximity and focal mechanism features are important. [ABSTRACT FROM AUTHOR]