Machine Learning Emulation of Urban Land Surface Processes

Meyer, David E.; Grimmond, Sue; Dueben, Peter; Hogan, Robin J.; Reeuwijk, Maarten van

doi:10.1029/2021ms002744

Cited by 12 publications

(5 citation statements)

References 92 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Current research highlights challenges with NN emulators coupled to Earth system models, reporting degraded performance and unstable simulations under some circumstances (Brenowitz & Bretherton, 2019; Rasp et al., 2018). While our recent experience in emulating gravity wave drag (Chantry et al., 2021) and urban land surface (Meyer, Grimmond, et al., 2022) schemes was positive, long coupled evaluations are required to better assess these type of models for operational use.…”

Section: Discussionmentioning

confidence: 99%

Machine Learning Emulation of 3D Cloud Radiative Effects

Meyer

Hogan

Dueben

et al. 2022

J Adv Model Earth Syst

Self Cite

View full text Add to dashboard Cite

The treatment of cloud structure in numerical weather and climate models is often greatly simplified to make them computationally affordable. Here we propose to correct the European Centre for Medium‐Range Weather Forecasts 1D radiation scheme ecRad for 3D cloud effects using computationally cheap neural networks. 3D cloud effects are learned as the difference between ecRad's fast 1D Tripleclouds solver that neglects them and its 3D SPARTACUS (SPeedy Algorithm for Radiative TrAnsfer through CloUd Sides) solver that includes them but is about five times more computationally expensive. With typical errors between 20% and 30% of the 3D signal, neural networks improve Tripleclouds' accuracy for about 1% increase in runtime. Thus, rather than emulating the whole of SPARTACUS, we keep Tripleclouds unchanged for cloud‐free parts of the atmosphere and 3D‐correct it elsewhere. The focus on the comparably small 3D correction instead of the entire signal allows us to improve predictions significantly if we assume a similar signal‐to‐noise ratio for both.

show abstract

Section: Discussionmentioning

confidence: 99%

Machine Learning Emulation of 3D Cloud Radiative Effects

Meyer

Hogan

Dueben

et al. 2022

J Adv Model Earth Syst

Self Cite

View full text Add to dashboard Cite

show abstract

“…The obtained refinement using simulations such as LES can illustrate how mitigation strategies would impact urban comfort, as discussed in [134]. The combinations will further allow predicting local extreme meteorological events more accurately [135].…”

Section: Monitoring Climate Variabilitymentioning

confidence: 95%

A review on how Big Data can help to monitor the environment and to mitigate risks due to climate change

Montillet,

Kermarrec,

Forootan

et al. 2024

Preprint

View full text Add to dashboard Cite

Climate change triggers a wide range of hydrometeorological, glaciological and geophysical processes that span across vast spatiotemporal scales. With the advances in technology and analytics, a multitude of remote sensing, geodetic and in situ instruments have been developed to effectively monitor and help comprehend the Earth's system including its climate variability and the recent anomalies associated with global warming. A huge volume of data is generated by recording these observations, resulting in the need for novel methods to handle and interpret such Big Datasets. Managing this enormous amount of data extends beyond current computer storage considerations; it also encompasses the complexities of processing, modeling, and analysing. Big Datasets present unique characteristics that set them apart from smaller datasets, thereby posing challenges to traditional approaches. Moreover, computational time plays a crucial role, especially in the context of geohazard warning and response systems which necessitate low latency requirements. In this review, we delve into the monitoring and analysis of various climate change-related phenomena, including, but not limited to, droughts, floods, cyclones-induced storm surges, urban heat islands, ice mass balance, sea-level rise, and the modelling of the influence of solar variability on the Earth’s climate. By examining these phenomena, we explore some of the current and future trends in Big Data, aiming to encourage and speed-up the development of such techniques and promoting their benefits to timely monitor and towards achieving climate sustainability, thereby addressing its threat to humanity.

show abstract

“…In SpeedyWeather.jl, interfaces to the model components are exposed to the user. Furthermore, data-driven climate modelling (Rasp et al, 2018;Schneider et al, 2023), which replaces existing model components with machine learning, is more difficult in Fortran due to the lack of established machine learning frameworks (Meyer et al, 2022). In Julia, Flux.jl (Innes et al, 2019) is available for machine learning as well as automatic differentiation with Enzyme (Moses & Churavy, 2020) for gradients-based optimization.…”

Section: Statement Of Needmentioning

confidence: 99%