Opinion: Optimizing climate models with process knowledge, resolution, and artificial intelligence

Atmospheric chemistry and physics, Journal Year: 2024, Volume and Issue: 24(12), P. 7041 - 7062

Published: June 19, 2024

Abstract. Accelerated progress in climate modeling is urgently needed for proactive and effective change adaptation. The central challenge lies accurately representing processes that are small scale yet climatically important, such as turbulence cloud formation. These will not be explicitly resolvable the foreseeable future, necessitating use of parameterizations. We propose a balanced approach leverages strengths traditional process-based parameterizations contemporary artificial intelligence (AI)-based methods to model subgrid-scale processes. This strategy employs AI derive data-driven closure functions from both observational simulated data, integrated within encode system knowledge conservation laws. In addition, increasing resolution resolve larger fraction small-scale can aid toward improved interpretable predictions outside observed distribution. However, currently feasible horizontal resolutions limited O(10 km) because higher would impede creation ensembles calibration uncertainty quantification, sampling atmospheric oceanic internal variability, broadly exploring quantifying risks. By synergizing decades scientific development with advanced techniques, our aims significantly boost accuracy, interpretability, trustworthiness predictions.

Language: Английский

---

Explainable AI: Bridging the Gap between Machine Learning Models and Human Understanding

Journal of Informatics Education and Research, Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 1, 2024

Explainable AI (XAI) is one of the key game-changing features in machine learning models, which contribute to making them more transparent, regulated and usable different applications. In (the) investigation this paper, we consider four rows explanation methods—LIME, SHAP, Anchor, Decision Tree-based Explanation—in disentangling decision-making process black box models within fields. our experiments, use datasets that cover domains, for example, health, finance image classification, compare accuracy, fidelity, coverage, precision human satisfaction each method. Our work shows rule trees approach called (Decision explanation) mostly superior comparison other non-model-specific methods performing higher coverage regardless classifier. addition this, respondents who answered qualitative evaluation indicated they were very content with decision tree-based explanations these types are easy understandable. Furthermore, most famous sorts clarifications instinctive significant. The over discoveries stretch on utilize interpretable strategies facilitating hole between understanding thus advancing straightforwardness responsibility AI-driven decision-making.

Language: Английский

---

Machine learning for the physics of climate

Nature Reviews Physics, Journal Year: 2024, Volume and Issue: 7(1), P. 6 - 20

Published: Nov. 11, 2024

Language: Английский

---

A rapid-application emissions-to-impacts tool for scenario assessment: Probabilistic Regional Impacts from Model patterns and Emissions (PRIME)

Geoscientific model development, Journal Year: 2025, Volume and Issue: 18(5), P. 1785 - 1808

Published: March 14, 2025

Abstract. Climate policies evolve quickly, and new scenarios designed around these are used to illustrate how they impact global mean temperatures using simple climate models (or emulators). Simple extremely efficient, although some can only provide estimates of metrics such as surface temperature, CO2 concentration effective radiative forcing. Within the Intergovernmental Panel on Change (IPCC) framework, understanding regional impacts that include most recent science is needed allow targeted policy decisions be made quickly. To address this, we present PRIME (Probabilistic Regional Impacts from Model patterns Emissions), a flexible probabilistic framework which aims an efficient mechanism run without significant overheads larger, more complex Earth system (ESMs). provides capability features ESM projections, ensemble simulations multi-centennial timescales analyses many key variables relevant important for assessments. We use model temperature response emissions scenarios. These estimated scale monthly large number CMIP6 ESMs. inputs “weather generator” algorithm land model. The thus generates end-to-end estimate test known in form shared socioeconomic pathways (SSPs), demonstrate our reproduces responses show results range scenarios: SSP5–8.5 high-emissions scenario was define patterns, SSP1–2.6, mitigation with low emissions, SSP5–3.4-OS, overshoot scenario, were verification data. correctly represents (and spread) scenarios, gives us confidence simulation will useful rapidly providing spatially resolved information novel thereby substantially reducing time between being released availability information.

Language: Английский

---

Gravity Wave Momentum Fluxes from 1 km Global ECMWF Integrated Forecast System

Scientific Data, Journal Year: 2024, Volume and Issue: 11(1)

Published: Aug. 21, 2024

Progress in understanding the impact of mesoscale variability, including gravity waves (GWs), on atmospheric circulation is often limited by availability global fine-resolution observations and simulated data. This study presents momentum fluxes due to GWs extracted from four months an experimental "nature run", integrated at a 1 km resolution (XNR1K) using Integrated Forecast System (IFS) model. Helmholtz decomposition used compute zonal meridional flux vertical ~1.5 petabytes data; quantities emulated climate model parameterization GWs. The are validated ERA5 reanalysis, both during first week after initialization over boreal winter period November 2018 February 2019. agreement between reanalysis IFS demonstrates its capability generate reliable distributions capture dynamic variability atmosphere. dataset could be valuable advancing our GW-planetary wave interactions, GW evolution around extremes, as high-quality training data for machine learning (ML) simulation

Language: Английский

---

Comparison of indicators to evaluate the performance of climate models

International Journal of Climatology, Journal Year: 2024, Volume and Issue: 44(13), P. 4907 - 4924

Published: Oct. 7, 2024

Abstract The evaluation of climate models is a crucial step in studies. It consists quantifying the resemblance model outputs to reference data identify with superior capacity replicate specific variables. Clearly, choice indicator significantly impacts results, underscoring importance selecting an that properly captures characteristics “good model”. This study examines behaviour six indicators, considering spatial correlation, distribution mean, variance and shape. Monthly for precipitation, temperature teleconnection patterns Central America were utilized analysis. A new multicomponent measure was selected based on these criteria assess performance 32 CMIP6 reproducing annual seasonal cycle top determined using multicriteria methods. found even best reproduces one derived climatic variable poorly this region. proposed selection method can contribute enhancing accuracy climatological research models.

Language: Английский

---