Harnessing the Power of Neural Operators with Automatically Encoded Conservation Laws

ArXiv 2312.11176 (2023)

Authors:

Ning Liu, Yiming Fan, Xianyi Zeng, Milan Klöwer, Lu Zhang, Yue Yu

A Machine Learning Approach for Predicting Essentiality of Metabolic Genes

In: Braman, J.C. (eds) Synthetic Biology. Methods in Molecular Biology, vol 2760 (2024)

Authors:

Lilli J Freischem & Diego A Oyarzún

Abstract:

The identification of essential genes is a key challenge in systems and synthetic biology, particularly for engineering metabolic pathways that convert feedstocks into valuable products. Assessment of gene essentiality at a genome scale requires large and costly growth assays of knockout strains. Here we describe a strategy to predict the essentiality of metabolic genes using binary classification algorithms. The approach combines elements from genome-scale metabolic models, directed graphs, and machine learning into a predictive model that can be trained on small knockout data. We demonstrate the efficacy of this approach using the most complete metabolic model of Escherichia coli and various machine learning algorithms for binary classification.

Neural General Circulation Models for Weather and Climate

(2023)

Authors:

Dmitrii Kochkov, Janni Yuval, Ian Langmore, Peter Norgaard, Jamie Smith, Griffin Mooers, Milan Klöwer, James Lottes, Stephan Rasp, Peter Düben, Sam Hatfield, Peter Battaglia, Alvaro Sanchez-Gonzalez, Matthew Willson, Michael P Brenner, Stephan Hoyer

Rapid saturation of cloud water adjustments to shipping emissions

Atmospheric Chemistry and Physics European Geosciences Union 23:19 (2023) 12545-12555

Authors:

Peter Manshausen, Duncan Watson-Parris, Matthew W Christensen, Jukka-Pekka Jalkanen, Philip Stier

Abstract:

Human aerosol emissions change cloud properties by providing additional cloud condensation nuclei. This increases cloud droplet numbers, which in turn affects other cloud properties like liquid-water content and ultimately cloud albedo. These adjustments are poorly constrained, making aerosol effects the most uncertain part of anthropogenic climate forcing. Here we show that cloud droplet number and water content react differently to changing emission amounts in shipping exhausts. We use information about ship positions and modeled emission amounts together with reanalysis winds and satellite retrievals of cloud properties. The analysis reveals that cloud droplet numbers respond linearly to emission amount over a large range (1–10 kg h−1) before the response saturates. Liquid water increases in raining clouds, and the anomalies are constant over the emission ranges observed. There is evidence that this independence of emissions is due to compensating effects under drier and more humid conditions, consistent with suppression of rain by enhanced aerosol. This has implications for our understanding of cloud processes and may improve the way clouds are represented in climate models, in particular by changing parameterizations of liquid-water responses to aerosol.

A Lagrangian perspective on the lifecycle and cloud radiative effect of deep convective clouds over Africa

EGUsphere (2023)

Authors:

William K Jones, Martin Stengel, Philip Stier

Abstract:

The anvil clouds of tropical deep convection have large radiative effects in both the shortwave (SW) and longwave (LW) spectra with the average magnitudes of both over 100 Wm-2. Despite this, due to the opposite sign of these fluxes, the net average of anvil cloud radiative effect (CRE) over the tropics has been found to be neutral. Research into the response of anvil CRE to climate change has primarily focused on the feedbacks of anvil cloud height and anvil cloud area, in particular regarding the LW feedback. However, tropical deep convection over land has a strong diurnal cycle which may couple with the shortwave component of anvil cloud radiative effect. As this diurnal cycle is poorly represented in climate models it is vital to gain a better understanding of how its changes impact anvil CRE.


To study the connection between deep convective cloud (DCC) lifecycle and CRE, we investigate the behaviour of both isolated and organised DCCs in a 4-month case study over sub-Saharan Africa (May–August 2016). Using a novel cloud tracking algorithm, we detect and track growing convective cores and their associated anvil clouds using geostationary satellite observations from Meteosat SEVIRI. Retrieved cloud properties and derived broadband radiative fluxes are provided by the CC4CL algorithm. By collecting the cloud properties of the tracked DCCs, we produce a dataset of anvil cloud properties along their lifetimes. While the majority of DCCs tracked in this dataset are isolated, with only a single core, the overall coverage of anvil clouds is dominated by those of clustered, multi-core anvils due to their larger areas and lifetimes.


We find that the distribution of anvil cloud CRE of our tracked DCCs has a bimodal distribution. The interaction between the lifecycles of DCCs and the diurnal cycle of insolation results in a wide range of SW anvil CRE, while the LW component remains in a comparatively narrow range of values. The CRE of individual anvil clouds varies widely, with isolated DCCs tending to have large negative or positive CREs while larger, organised systems tend to have CRE closer to zero. Despite this, we find that the net anvil cloud CRE across all tracked DCCs is indeed neutral within our range of uncertainty (0.86 ± 0.91 Wm-2). Changes in the lifecycle of DCCs, such as shifts in the time of triggering, or the length of the dissipating phase, could have large impacts on the SW anvil CRE and lead to complex responses that are not considered by theories of LW anvil CRE feedbacks.