Search for: All records

Accurate and computationally-viable representations of clouds and turbulence are a long-standing challenge for climate model development. Traditional parameterizations that crudely but efficiently approximate these processes are a leading source of uncertainty in long-term projected warming and precipitation patterns. Machine Learning (ML)-based parameterizations have long been hailed as a promising alternative with the potential to yield higher accuracy at a fraction of the cost of more explicit simulations. However, these ML variants are often unpredictably unstable and inaccurate in \textit{coupled} testing (i.e. in a downstream hybrid simulation task where they are dynamically interacting with the large-scale climate model). These issues are exacerbated in out-of-distribution climates. Certain design decisions such as ``climate-invariant" feature transformation for moisture inputs, input vector expansion, and temporal history incorporation have been shown to improve coupled performance, but they may be insufficient for coupled out-of-distribution generalization. If feature selection and transformations can inoculate hybrid physics-ML climate models from non-physical, out-of-distribution extrapolation in a changing climate, there is far greater potential in extrapolating from observational data. Otherwise, training on multiple simulated climates becomes an inevitable necessity. While our results show generalization benefits from these design decisions, the obtained improvment does not sufficiently preclude the necessity of using multi-climate simulated training data. 

Abstract Enigmatic large‐scale (>150 km wide) ground deformation in southern Bolivia has been ongoing for more than 50 year. Concurrent changes in gravity recorded between 2010 and 2018 imply minor changes in subsurface density in the absence of significant mass changes. Numerical modeling of the gravity changes and concurrent InSAR LOS displacements gives annual bulk density changes of 0.002 kg m⁻³in the Altiplano‐Puna Magma Body (APMB) and −0.03 kg m⁻³in a vertical bulge‐column ensemble beneath Uturuncu volcano. We propose that the transcrustal migration of fluids from the APMB to shallower crustal levels by compressible flow is the source of ground deformation. Localized ground subsidence south of Uturuncu can be best explained by a density decrease of 20 ± 5 kg m⁻³between 2011 and 2013 in a hydrothermal reservoir. Our findings contribute to the growing recognition of transcrustal fluid migration as a source of volcanic unrest. 

Modern climate projections lack adequate spatial and temporal resolution due to computational constraints. A consequence is inaccurate and imprecise predictions of critical processes such as storms. Hybrid methods that combine physics with machine learning (ML) have introduced a new generation of higher fidelity climate simulators that can sidestep Moore's Law by outsourcing compute-hungry, short, high-resolution simulations to ML emulators. However, this hybrid ML-physics simulation approach requires domain-specific treatment and has been inaccessible to ML experts because of lack of training data and relevant, easy-to-use workflows. We present ClimSim, the largest-ever dataset designed for hybrid ML-physics research. It comprises multi-scale climate simulations, developed by a consortium of climate scientists and ML researchers. It consists of 5.7 billion pairs of multivariate input and output vectors that isolate the influence of locally-nested, high-resolution, high-fidelity physics on a host climate simulator's macro-scale physical state.The dataset is global in coverage, spans multiple years at high sampling frequency, and is designed such that resulting emulators are compatible with downstream coupling into operational climate simulators. We implement a range of deterministic and stochastic regression baselines to highlight the ML challenges and their scoring. The data (https://huggingface.co/datasets/LEAP/ClimSim_high-res) and code (https://leap-stc.github.io/ClimSim) are released openly to support the development of hybrid ML-physics and high-fidelity climate simulations for the benefit of science and society. 

Abstract The Madden–Julian Oscillation (MJO) is widely acknowledged for its ability to modulate Northwest Pacific tropical cyclones (TCs), but a complete understanding of the underlying mechanisms remains uncertain. Beyond established effects of the MJO's relative humidity envelope, other dynamical factors have recently been invoked via new genesis potential indices and high‐resolution modeling studies. Here we revisit the ability of the MJO to modulate West Pacific TCs through a quasi‐explicit cyclone downscaling strategy driven by composited observations, paired later with a genesis index to investigate regional drivers of modulation. We reveal two distinct spatial modes of TC modulation in which the MJO's dynamic and thermodynamic effects act in tandem to increase TCs. In the South China Sea, for instance, shear reductions associated with the MJO's circulation lead to increasing potential intensity ahead of the arrival of a positive humidity anomaly, all of which combine for an extended period of cyclogenesis favorability. 

ABSTRACT Regional climate modeling addresses our need to understand and simulate climatic processes and phenomena unresolved in global models. This paper highlights examples of current approaches to and innovative uses of regional climate modeling that deepen understanding of the climate system. High-resolution models are generally more skillful in simulating extremes, such as heavy precipitation, strong winds, and severe storms. In addition, research has shown that fine-scale features such as mountains, coastlines, lakes, irrigation, land use, and urban heat islands can substantially influence a region’s climate and its response to changing forcings. Regional climate simulations explicitly simulating convection are now being performed, providing an opportunity to illuminate new physical behavior that previously was represented by parameterizations with large uncertainties. Regional and global models are both advancing toward higher resolution, as computational capacity increases. However, the resolution and ensemble size necessary to produce a sufficient statistical sample of these processes in global models has proven too costly for contemporary supercomputing systems. Regional climate models are thus indispensable tools that complement global models for understanding physical processes governing regional climate variability and change. The deeper understanding of regional climate processes also benefits stakeholders and policymakers who need physically robust, high-resolution climate information to guide societal responses to changing climate. Key scientific questions that will continue to require regional climate models, and opportunities are emerging for addressing those questions.