Francesco Fedele, Cristel Chandre, Martin Horvat, Nedjeljka Žagar
The reduced-complexity models developed by Edward Lorenz are widely used in atmospheric and climate sciences to study nonlinear aspect of dynamics and to demonstrate new methods for numerical weather prediction. A set of inviscid Lorenz models describing the dynamics of a single variable in a zonally-periodic domain, without dissipation and forcing, conserve energy but are not Hamiltonian. In this paper, we start from a general continuous parent fluid model, from which we derive a family of Hamiltonian Lorenz-like models through a symplectic discretization of the associated Poisson bracket that preserves the Jacobi identity. A symplectic-split integrator is also formulated. These Hamiltonian models conserve energy and maintain the nearest-neighbor couplings inherent in the original Lorenz model. As a corollary, we find that the Lorenz-96 model can be seen as a result of a poor discretization of a Poisson bracket. Hamiltonian Lorenz-like models offer promising alternatives to the original Lorenz models, especially for the qualitative representation of non-Gaussian weather extremes and wave interactions, which are key factors in understanding many phenomena of the climate system.
{"title":"Hamiltonian Lorenz-like models","authors":"Francesco Fedele, Cristel Chandre, Martin Horvat, Nedjeljka Žagar","doi":"arxiv-2409.07920","DOIUrl":"https://doi.org/arxiv-2409.07920","url":null,"abstract":"The reduced-complexity models developed by Edward Lorenz are widely used in\u0000atmospheric and climate sciences to study nonlinear aspect of dynamics and to\u0000demonstrate new methods for numerical weather prediction. A set of inviscid\u0000Lorenz models describing the dynamics of a single variable in a\u0000zonally-periodic domain, without dissipation and forcing, conserve energy but\u0000are not Hamiltonian. In this paper, we start from a general continuous parent\u0000fluid model, from which we derive a family of Hamiltonian Lorenz-like models\u0000through a symplectic discretization of the associated Poisson bracket that\u0000preserves the Jacobi identity. A symplectic-split integrator is also\u0000formulated. These Hamiltonian models conserve energy and maintain the\u0000nearest-neighbor couplings inherent in the original Lorenz model. As a\u0000corollary, we find that the Lorenz-96 model can be seen as a result of a poor\u0000discretization of a Poisson bracket. Hamiltonian Lorenz-like models offer\u0000promising alternatives to the original Lorenz models, especially for the\u0000qualitative representation of non-Gaussian weather extremes and wave\u0000interactions, which are key factors in understanding many phenomena of the\u0000climate system.","PeriodicalId":501166,"journal":{"name":"arXiv - PHYS - Atmospheric and Oceanic Physics","volume":"28 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142215327","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Snow is a crucial element of the sea ice system, affecting sea ice growth and decay due to its low thermal conductivity and high albedo. Despite its importance, present-day climate models have an idealized representation of snow, often including only single-layer thermodynamics and omitting several processes that shape its properties. Although advanced snow process models like SnowModel exist, they are often excluded from climate modeling due to their high computational costs. SnowModel simulates snow depth, density, blowing-snow redistribution, sublimation, grain size, and thermal conductivity in a multi-layer snowpack. It operates with high spatial (1 meter) and temporal (1 hour) resolution. However, for large regions like the Arctic Ocean, these high-resolution simulations face challenges such as slow processing and large resource requirements. Data-driven emulators are used to address these issues, but they often lack generalizability and consistency with physical laws. In our study, we address these challenges by developing a physics-guided emulator that incorporates physical laws governing changes in snow density due to compaction. We evaluated three machine learning models: Long Short-Term Memory (LSTM), Physics-Guided LSTM, and Random Forest across five Arctic regions. All models achieved high accuracy, with the Physics-Guided LSTM showing the best performance in accuracy and generalizability. Our approach offers a faster way to emulate SnowModel with a speedup of over 9000 times, maintaining high fidelity.
{"title":"Modeling Snow on Sea Ice using Physics Guided Machine Learning","authors":"Ayush Prasad, Ioanna Merkouriadi, Aleksi Nummelin","doi":"arxiv-2409.08092","DOIUrl":"https://doi.org/arxiv-2409.08092","url":null,"abstract":"Snow is a crucial element of the sea ice system, affecting sea ice growth and\u0000decay due to its low thermal conductivity and high albedo. Despite its\u0000importance, present-day climate models have an idealized representation of\u0000snow, often including only single-layer thermodynamics and omitting several\u0000processes that shape its properties. Although advanced snow process models like\u0000SnowModel exist, they are often excluded from climate modeling due to their\u0000high computational costs. SnowModel simulates snow depth, density, blowing-snow\u0000redistribution, sublimation, grain size, and thermal conductivity in a\u0000multi-layer snowpack. It operates with high spatial (1 meter) and temporal (1\u0000hour) resolution. However, for large regions like the Arctic Ocean, these\u0000high-resolution simulations face challenges such as slow processing and large\u0000resource requirements. Data-driven emulators are used to address these issues,\u0000but they often lack generalizability and consistency with physical laws. In our\u0000study, we address these challenges by developing a physics-guided emulator that\u0000incorporates physical laws governing changes in snow density due to compaction.\u0000We evaluated three machine learning models: Long Short-Term Memory (LSTM),\u0000Physics-Guided LSTM, and Random Forest across five Arctic regions. All models\u0000achieved high accuracy, with the Physics-Guided LSTM showing the best\u0000performance in accuracy and generalizability. Our approach offers a faster way\u0000to emulate SnowModel with a speedup of over 9000 times, maintaining high\u0000fidelity.","PeriodicalId":501166,"journal":{"name":"arXiv - PHYS - Atmospheric and Oceanic Physics","volume":"2 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142215329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
It is largely understood that subseasonal-to-seasonal (S2S) predictability arises from the atmospheric initial state during early lead times, the land during intermediate lead times, and the ocean during later lead times. We examine whether this hypothesis holds for the S2S prediction of weather regimes by training a set of XGBoost models to predict weekly weather regimes over North America at 1-to-8-week lead times. Each model used a different predictor from one of the three considered Earth system components (atmosphere, ocean, or land) sourced from reanalyses. Three additional models were trained using land-, ocean-, or atmosphere-only predictors to capture process interactions and leverage multiple signals within the respective Earth system component. We found that each Earth system component performed more skillfully at different forecast horizons, with sensitivity to seasonality and observed (i.e., ground truth) weather regime. S2S predictability from the atmosphere was higher during winter, from the ocean during summer, and from land during spring and summer. Ocean heat content was the best predictor for most seasons and weather regimes beyond week 2, highlighting the importance of sub-surface ocean conditions for S2S predictability. Soil temperature and water content were also important predictors. Climate patterns were associated with changes in the likelihood of occurrence for specific weather regimes, including the El Ni~no-Southern Oscillation, Madden Julian Oscillation, North Pacific Gyre, and Indian Ocean dipole. This study quantifies predictability from some previously identified processes on the large-scale atmospheric circulation and gives insight into new sources for future study.
{"title":"An Earth-System-Oriented View of the S2S Predictability of North American Weather Regimes","authors":"Jhayron S. Pérez-Carrasquilla, Maria J. Molina","doi":"arxiv-2409.08174","DOIUrl":"https://doi.org/arxiv-2409.08174","url":null,"abstract":"It is largely understood that subseasonal-to-seasonal (S2S) predictability\u0000arises from the atmospheric initial state during early lead times, the land\u0000during intermediate lead times, and the ocean during later lead times. We\u0000examine whether this hypothesis holds for the S2S prediction of weather regimes\u0000by training a set of XGBoost models to predict weekly weather regimes over\u0000North America at 1-to-8-week lead times. Each model used a different predictor\u0000from one of the three considered Earth system components (atmosphere, ocean, or\u0000land) sourced from reanalyses. Three additional models were trained using\u0000land-, ocean-, or atmosphere-only predictors to capture process interactions\u0000and leverage multiple signals within the respective Earth system component. We\u0000found that each Earth system component performed more skillfully at different\u0000forecast horizons, with sensitivity to seasonality and observed (i.e., ground\u0000truth) weather regime. S2S predictability from the atmosphere was higher during\u0000winter, from the ocean during summer, and from land during spring and summer.\u0000Ocean heat content was the best predictor for most seasons and weather regimes\u0000beyond week 2, highlighting the importance of sub-surface ocean conditions for\u0000S2S predictability. Soil temperature and water content were also important\u0000predictors. Climate patterns were associated with changes in the likelihood of\u0000occurrence for specific weather regimes, including the El Ni~no-Southern\u0000Oscillation, Madden Julian Oscillation, North Pacific Gyre, and Indian Ocean\u0000dipole. This study quantifies predictability from some previously identified\u0000processes on the large-scale atmospheric circulation and gives insight into new\u0000sources for future study.","PeriodicalId":501166,"journal":{"name":"arXiv - PHYS - Atmospheric and Oceanic Physics","volume":"38 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142215326","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Fitting nonlinear dynamical models to sparse and noisy observations is fundamentally challenging. Identifying dynamics requires data assimilation (DA) to estimate system states, but DA requires an accurate dynamical model. To break this deadlock we present CODA, an end-to-end optimization scheme for jointly learning dynamics and DA directly from sparse and noisy observations. A neural network is trained to carry out data accurate, efficient and parallel-in-time DA, while free parameters of the dynamical system are simultaneously optimized. We carry out end-to-end learning directly on observation data, introducing a novel learning objective that combines unrolled auto-regressive dynamics with the data- and self-consistency terms of weak-constraint 4Dvar DA. By taking into account interactions between new and existing simulation components over multiple time steps, CODA can recover initial conditions, fit unknown dynamical parameters and learn neural network-based PDE terms to match both available observations and self-consistency constraints. In addition to facilitating end-to-end learning of dynamics and providing fast, amortized, non-sequential DA, CODA provides greater robustness to model misspecification than classical DA approaches.
将非线性动力学模型拟合到稀疏且高噪声的观测数据中是一项艰巨的任务。识别动力学需要数据同化（DA）来估计系统状态，但 DA 需要精确的动力学模型。为了打破这一僵局，我们提出了 CODA，这是一种端到端优化方案，可直接从稀疏和嘈杂的观测数据中联合学习动力学和数据同化。训练神经网络以进行数据精确、高效和并行的实时数据分析，同时优化动力学系统的自由参数。我们直接在观测数据上进行端到端学习，引入了一种新的学习目标，它将非滚动自回归动力学与弱约束 4Dvar DA 的数据和自一致性条款相结合。通过考虑多个时间步长中新的和现有的模拟组件之间的相互作用，CODA 可以恢复初始条件、拟合未知的动力学参数并学习基于神经网络的 PDE 项，以匹配可用观测数据和自一致性约束。除了促进端到端动力学学习和提供快速、摊销式、非序列 DA 之外，CODA 还提供了比经典 DA 方法更强的模型错误规范鲁棒性。
{"title":"Combined Optimization of Dynamics and Assimilation with End-to-End Learning on Sparse Observations","authors":"Vadim Zinchenko, David S. Greenberg","doi":"arxiv-2409.07137","DOIUrl":"https://doi.org/arxiv-2409.07137","url":null,"abstract":"Fitting nonlinear dynamical models to sparse and noisy observations is\u0000fundamentally challenging. Identifying dynamics requires data assimilation (DA)\u0000to estimate system states, but DA requires an accurate dynamical model. To\u0000break this deadlock we present CODA, an end-to-end optimization scheme for\u0000jointly learning dynamics and DA directly from sparse and noisy observations. A\u0000neural network is trained to carry out data accurate, efficient and\u0000parallel-in-time DA, while free parameters of the dynamical system are\u0000simultaneously optimized. We carry out end-to-end learning directly on\u0000observation data, introducing a novel learning objective that combines unrolled\u0000auto-regressive dynamics with the data- and self-consistency terms of\u0000weak-constraint 4Dvar DA. By taking into account interactions between new and\u0000existing simulation components over multiple time steps, CODA can recover\u0000initial conditions, fit unknown dynamical parameters and learn neural\u0000network-based PDE terms to match both available observations and\u0000self-consistency constraints. In addition to facilitating end-to-end learning\u0000of dynamics and providing fast, amortized, non-sequential DA, CODA provides\u0000greater robustness to model misspecification than classical DA approaches.","PeriodicalId":501166,"journal":{"name":"arXiv - PHYS - Atmospheric and Oceanic Physics","volume":"42 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142215368","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Muhammad Akhtar Munir, Fahad Shahbaz Khan, Salman Khan
Accurate weather and climate modeling is critical for both scientific advancement and safeguarding communities against environmental risks. Traditional approaches rely heavily on Numerical Weather Prediction (NWP) models, which simulate energy and matter flow across Earth's systems. However, heavy computational requirements and low efficiency restrict the suitability of NWP, leading to a pressing need for enhanced modeling techniques. Neural network-based models have emerged as promising alternatives, leveraging data-driven approaches to forecast atmospheric variables. In this work, we focus on limited-area modeling and train our model specifically for localized region-level downstream tasks. As a case study, we consider the MENA region due to its unique climatic challenges, where accurate localized weather forecasting is crucial for managing water resources, agriculture and mitigating the impacts of extreme weather events. This targeted approach allows us to tailor the model's capabilities to the unique conditions of the region of interest. Our study aims to validate the effectiveness of integrating parameter-efficient fine-tuning (PEFT) methodologies, specifically Low-Rank Adaptation (LoRA) and its variants, to enhance forecast accuracy, as well as training speed, computational resource utilization, and memory efficiency in weather and climate modeling for specific regions.
{"title":"Efficient Localized Adaptation of Neural Weather Forecasting: A Case Study in the MENA Region","authors":"Muhammad Akhtar Munir, Fahad Shahbaz Khan, Salman Khan","doi":"arxiv-2409.07585","DOIUrl":"https://doi.org/arxiv-2409.07585","url":null,"abstract":"Accurate weather and climate modeling is critical for both scientific\u0000advancement and safeguarding communities against environmental risks.\u0000Traditional approaches rely heavily on Numerical Weather Prediction (NWP)\u0000models, which simulate energy and matter flow across Earth's systems. However,\u0000heavy computational requirements and low efficiency restrict the suitability of\u0000NWP, leading to a pressing need for enhanced modeling techniques. Neural\u0000network-based models have emerged as promising alternatives, leveraging\u0000data-driven approaches to forecast atmospheric variables. In this work, we\u0000focus on limited-area modeling and train our model specifically for localized\u0000region-level downstream tasks. As a case study, we consider the MENA region due\u0000to its unique climatic challenges, where accurate localized weather forecasting\u0000is crucial for managing water resources, agriculture and mitigating the impacts\u0000of extreme weather events. This targeted approach allows us to tailor the\u0000model's capabilities to the unique conditions of the region of interest. Our\u0000study aims to validate the effectiveness of integrating parameter-efficient\u0000fine-tuning (PEFT) methodologies, specifically Low-Rank Adaptation (LoRA) and\u0000its variants, to enhance forecast accuracy, as well as training speed,\u0000computational resource utilization, and memory efficiency in weather and\u0000climate modeling for specific regions.","PeriodicalId":501166,"journal":{"name":"arXiv - PHYS - Atmospheric and Oceanic Physics","volume":"4 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142227032","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Xiaohui Zhong, Lei Chen, Xu Fan, Wenxu Qian, Jun Liu, Hao Li
Machine learning (ML) models have become increasingly valuable in weather forecasting, providing forecasts that not only lower computational costs but often match or exceed the accuracy of traditional numerical weather prediction (NWP) models. Despite their potential, ML models typically suffer from limitations such as coarse temporal resolution, typically 6 hours, and a limited set of meteorological variables, limiting their practical applicability. To overcome these challenges, we introduce FuXi-2.0, an advanced ML model that delivers 1-hourly global weather forecasts and includes a comprehensive set of essential meteorological variables, thereby expanding its utility across various sectors like wind and solar energy, aviation, and marine shipping. Our study conducts comparative analyses between ML-based 1-hourly forecasts and those from the high-resolution forecast (HRES) of the European Centre for Medium-Range Weather Forecasts (ECMWF) for various practical scenarios. The results demonstrate that FuXi-2.0 consistently outperforms ECMWF HRES in forecasting key meteorological variables relevant to these sectors. In particular, FuXi-2.0 shows superior performance in wind power forecasting compared to ECMWF HRES, further validating its efficacy as a reliable tool for scenarios demanding precise weather forecasts. Additionally, FuXi-2.0 also integrates both atmospheric and oceanic components, representing a significant step forward in the development of coupled atmospheric-ocean models. Further comparative analyses reveal that FuXi-2.0 provides more accurate forecasts of tropical cyclone intensity than its predecessor, FuXi-1.0, suggesting that there are benefits of an atmosphere-ocean coupled model over atmosphere-only models.
机器学习（ML）模型在天气预报中的价值与日俱增，它提供的预报不仅能降低计算成本，而且精度往往能达到或超过传统的数值天气预报（NWP）模型。尽管 ML 模型潜力巨大，但它通常受到时间分辨率较低（通常为 6 小时）和气象变量集有限等限制，从而限制了其实际应用性。为了克服这些挑战，我们引入了 FuXi-2.0，这是一种先进的 ML 模式，可提供每 1 小时的全球天气预报，并包含一套全面的基本气象变量，从而将其用途扩展到风能和太阳能、航空和海运等各个领域。我们的研究对基于 ML 的 1 小时预报和欧洲中期天气预报中心（ECMWF）的高分辨率预报（HRES）进行了比较分析。结果表明，FuXi-2.0 在预报与这些部门相关的关键气象变量方面始终优于 ECMWFHRES。特别是，与 ECMWF HRES 相比，FuXi-2.0 在风力发电预报方面表现出更优越的性能，进一步验证了其作为需要精确天气预报的情况下的可靠工具的有效性。此外，FuXi-2.0 还集成了大气和海洋成分，在开发大气-海洋耦合模式方面迈出了重要一步。进一步的比较分析表明，FuXi-2.0 对热带气旋强度的预报比其前身 FuXi-1.0 更准确，这表明大气-海洋耦合模式比纯大气模式更有优势。
{"title":"FuXi-2.0: Advancing machine learning weather forecasting model for practical applications","authors":"Xiaohui Zhong, Lei Chen, Xu Fan, Wenxu Qian, Jun Liu, Hao Li","doi":"arxiv-2409.07188","DOIUrl":"https://doi.org/arxiv-2409.07188","url":null,"abstract":"Machine learning (ML) models have become increasingly valuable in weather\u0000forecasting, providing forecasts that not only lower computational costs but\u0000often match or exceed the accuracy of traditional numerical weather prediction\u0000(NWP) models. Despite their potential, ML models typically suffer from\u0000limitations such as coarse temporal resolution, typically 6 hours, and a\u0000limited set of meteorological variables, limiting their practical\u0000applicability. To overcome these challenges, we introduce FuXi-2.0, an advanced\u0000ML model that delivers 1-hourly global weather forecasts and includes a\u0000comprehensive set of essential meteorological variables, thereby expanding its\u0000utility across various sectors like wind and solar energy, aviation, and marine\u0000shipping. Our study conducts comparative analyses between ML-based 1-hourly\u0000forecasts and those from the high-resolution forecast (HRES) of the European\u0000Centre for Medium-Range Weather Forecasts (ECMWF) for various practical\u0000scenarios. The results demonstrate that FuXi-2.0 consistently outperforms ECMWF\u0000HRES in forecasting key meteorological variables relevant to these sectors. In\u0000particular, FuXi-2.0 shows superior performance in wind power forecasting\u0000compared to ECMWF HRES, further validating its efficacy as a reliable tool for\u0000scenarios demanding precise weather forecasts. Additionally, FuXi-2.0 also\u0000integrates both atmospheric and oceanic components, representing a significant\u0000step forward in the development of coupled atmospheric-ocean models. Further\u0000comparative analyses reveal that FuXi-2.0 provides more accurate forecasts of\u0000tropical cyclone intensity than its predecessor, FuXi-1.0, suggesting that\u0000there are benefits of an atmosphere-ocean coupled model over atmosphere-only\u0000models.","PeriodicalId":501166,"journal":{"name":"arXiv - PHYS - Atmospheric and Oceanic Physics","volume":"281 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142215366","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In recent years, extreme weather events have surged, highlighting the urgent need for action on the climate emergency. The year 2023 saw record-breaking global temperatures, unprecedented heatwaves in Europe, devastating floods in Asia, and severe wildfires in North America and Australia. Super El Ni~no events, known for their profound impact on global weather, play a critical role in these changes, causing severe economic and environmental damage. This study presents a novel predictive model that integrates systematically local and global climatic signals to forecast Super El Ni~no events, introducing the Super El Ni~no Index (SEI), which value of 80 or higher defines a Super El Ni~no event. Our analysis shows that the SEI accurately reflects past Super El Ni~no events, including those from 1982-83, 1997-98, and 2015-16, with SEI values for these periods containing 80 within the 2-sigma standard deviation. Using data up to 2022, our model predicted an SEI of around 80 for 2023, indicating a Super El Ni~no for the 2023-24 period. Recent observations confirm that the 2023-24 El Ni~no is among the five strongest recorded Super El Ni~no events in history. An analysis of SEI trends from 1982 to 2023 reveals a gradual increase, with recent El Ni~no events consistently exceeding SEI values of 70. This trend suggests that El Ni~no events are increasingly approaching Super El Ni~no intensity, potentially due to more favorable conditions in the equatorial Pacific. This increase in SEI values and the frequency of stronger El Ni~no events may be attributed to the ongoing effects of global warming. These findings emphasize the need for heightened preparedness and strategic planning to mitigate the impacts of future Super El Ni~no events, which are likely to become more frequent in the coming decades.
{"title":"Deciphering Super El Niño: Development of a Novel Predictive Model Integrating Local and Global Climatic Signals","authors":"Chae-Hyun Yoon, Jubin Park, Myung-Ki Cheoun","doi":"arxiv-2409.06161","DOIUrl":"https://doi.org/arxiv-2409.06161","url":null,"abstract":"In recent years, extreme weather events have surged, highlighting the urgent\u0000need for action on the climate emergency. The year 2023 saw record-breaking\u0000global temperatures, unprecedented heatwaves in Europe, devastating floods in\u0000Asia, and severe wildfires in North America and Australia. Super El Ni~no\u0000events, known for their profound impact on global weather, play a critical role\u0000in these changes, causing severe economic and environmental damage. This study\u0000presents a novel predictive model that integrates systematically local and\u0000global climatic signals to forecast Super El Ni~no events, introducing the\u0000Super El Ni~no Index (SEI), which value of 80 or higher defines a Super El\u0000Ni~no event. Our analysis shows that the SEI accurately reflects past Super El\u0000Ni~no events, including those from 1982-83, 1997-98, and 2015-16, with SEI\u0000values for these periods containing 80 within the 2-sigma standard deviation.\u0000Using data up to 2022, our model predicted an SEI of around 80 for 2023,\u0000indicating a Super El Ni~no for the 2023-24 period. Recent observations\u0000confirm that the 2023-24 El Ni~no is among the five strongest recorded Super\u0000El Ni~no events in history. An analysis of SEI trends from 1982 to 2023\u0000reveals a gradual increase, with recent El Ni~no events consistently exceeding\u0000SEI values of 70. This trend suggests that El Ni~no events are increasingly\u0000approaching Super El Ni~no intensity, potentially due to more favorable\u0000conditions in the equatorial Pacific. This increase in SEI values and the\u0000frequency of stronger El Ni~no events may be attributed to the ongoing effects\u0000of global warming. These findings emphasize the need for heightened\u0000preparedness and strategic planning to mitigate the impacts of future Super El\u0000Ni~no events, which are likely to become more frequent in the coming decades.","PeriodicalId":501166,"journal":{"name":"arXiv - PHYS - Atmospheric and Oceanic Physics","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142215371","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jesse A. Miller, Merav Opher, Maria Hatzaki, Kyriakoula Papachristopoulou, Brian C. Thomas
Our solar system's path has recently been shown to potentially intersect dense interstellar clouds 2 and 7 million years ago: the Local Lynx of Cold Cloud and the edge of the Local Bubble. These clouds compressed the heliosphere, directly exposing Earth to the interstellar medium. Previous studies that examined climate effects of these encounters argued for an induced ice age due to the formation of global noctilucent clouds (NLCs). Here, we revisit such studies with a modern 2D atmospheric chemistry model using parameters of global heliospheric magnetohydrodynamic models as input. We show that NLCs remain confined to polar latitudes and short seasonal lifetimes during these dense cloud crossings lasting $sim10^5$ years. Polar mesospheric ozone becomes significantly depleted, but the total ozone column broadly increases. Furthermore, we show that the densest NLCs lessen the amount of sunlight reaching the surface instantaneously by up to 7% while halving outgoing longwave radiation.
{"title":"Earth's Mesosphere During Possible Encounters With Massive Interstellar Clouds 2 and 7 Million Years Ago","authors":"Jesse A. Miller, Merav Opher, Maria Hatzaki, Kyriakoula Papachristopoulou, Brian C. Thomas","doi":"arxiv-2409.06832","DOIUrl":"https://doi.org/arxiv-2409.06832","url":null,"abstract":"Our solar system's path has recently been shown to potentially intersect\u0000dense interstellar clouds 2 and 7 million years ago: the Local Lynx of Cold\u0000Cloud and the edge of the Local Bubble. These clouds compressed the\u0000heliosphere, directly exposing Earth to the interstellar medium. Previous\u0000studies that examined climate effects of these encounters argued for an induced\u0000ice age due to the formation of global noctilucent clouds (NLCs). Here, we\u0000revisit such studies with a modern 2D atmospheric chemistry model using\u0000parameters of global heliospheric magnetohydrodynamic models as input. We show\u0000that NLCs remain confined to polar latitudes and short seasonal lifetimes\u0000during these dense cloud crossings lasting $sim10^5$ years. Polar mesospheric\u0000ozone becomes significantly depleted, but the total ozone column broadly\u0000increases. Furthermore, we show that the densest NLCs lessen the amount of\u0000sunlight reaching the surface instantaneously by up to 7% while halving\u0000outgoing longwave radiation.","PeriodicalId":501166,"journal":{"name":"arXiv - PHYS - Atmospheric and Oceanic Physics","volume":"33 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142215373","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Giulio Passerotti, Alberto Alberello, Marcello Vichi, Luke G. Bennetts, Alessandro Toffoli
The size and shape of sea ice floes play a crucial role in influencing ocean-atmosphere energy exchanges, sea ice concentrations, albedo, and wave propagation through ice-covered waters. Despite the availability of diverse image segmentation techniques for analyzing sea ice imagery, accurately detecting and measuring floes remains a considerable challenge. This study presents a precise methodology for in-situ sea ice imagery acquisition, including automated orthorectification to correct perspective distortions. The image dataset, collected during an Antarctic winter expedition, was used to evaluate various automated image segmentation approaches: the traditional GVF Snake algorithm and the advanced deep learning model, Segment Anything Model (SAM). To address the limitations of each method, a hybrid algorithm combining traditional and AI-based techniques is proposed. The effectiveness of these approaches was validated through a detailed analysis of ice floe detection accuracy, floe size, and ice concentration statistics, with the outcomes normalized against a manually segmented benchmark.
{"title":"Segmenting sea ice floes in close-range optical imagery with active contour and foundation models","authors":"Giulio Passerotti, Alberto Alberello, Marcello Vichi, Luke G. Bennetts, Alessandro Toffoli","doi":"arxiv-2409.06641","DOIUrl":"https://doi.org/arxiv-2409.06641","url":null,"abstract":"The size and shape of sea ice floes play a crucial role in influencing\u0000ocean-atmosphere energy exchanges, sea ice concentrations, albedo, and wave\u0000propagation through ice-covered waters. Despite the availability of diverse\u0000image segmentation techniques for analyzing sea ice imagery, accurately\u0000detecting and measuring floes remains a considerable challenge. This study\u0000presents a precise methodology for in-situ sea ice imagery acquisition,\u0000including automated orthorectification to correct perspective distortions. The\u0000image dataset, collected during an Antarctic winter expedition, was used to\u0000evaluate various automated image segmentation approaches: the traditional GVF\u0000Snake algorithm and the advanced deep learning model, Segment Anything Model\u0000(SAM). To address the limitations of each method, a hybrid algorithm combining\u0000traditional and AI-based techniques is proposed. The effectiveness of these\u0000approaches was validated through a detailed analysis of ice floe detection\u0000accuracy, floe size, and ice concentration statistics, with the outcomes\u0000normalized against a manually segmented benchmark.","PeriodicalId":501166,"journal":{"name":"arXiv - PHYS - Atmospheric and Oceanic Physics","volume":"99 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142215369","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Longhao Wang, Xuanze Zhang, L. Ruby Leung, Francis H. S. Chiew, Amir AghaKouchak, Kairan Ying, Yongqiang Zhang
Accurate prediction of global sea surface temperature at sub-seasonal to seasonal (S2S) timescale is critical for drought and flood forecasting, as well as for improving disaster preparedness in human society. Government departments or academic studies normally use physics-based numerical models to predict S2S sea surface temperature and corresponding climate indices, such as El Ni~no-Southern Oscillation. However, these models are hampered by computational inefficiencies, limited retention of ocean-atmosphere initial conditions, and significant uncertainty and biases. Here, we introduce a novel three-dimensional deep learning neural network to model the nonlinear and complex coupled atmosphere-ocean weather systems. This model incorporates climatic and temporal features and employs a self-attention mechanism to enhance the prediction of global S2S sea surface temperature pattern. Compared to the physics-based models, it shows significant computational efficiency and predictive capability, improving one to three months sea surface temperature predictive skill by 13.7% to 77.1% in seven ocean regions with dominant influence on S2S variability over land. This achievement underscores the significant potential of deep learning for largely improving forecasting skills at the S2S scale over land.
{"title":"CAS-Canglong: A skillful 3D Transformer model for sub-seasonal to seasonal global sea surface temperature prediction","authors":"Longhao Wang, Xuanze Zhang, L. Ruby Leung, Francis H. S. Chiew, Amir AghaKouchak, Kairan Ying, Yongqiang Zhang","doi":"arxiv-2409.05369","DOIUrl":"https://doi.org/arxiv-2409.05369","url":null,"abstract":"Accurate prediction of global sea surface temperature at sub-seasonal to\u0000seasonal (S2S) timescale is critical for drought and flood forecasting, as well\u0000as for improving disaster preparedness in human society. Government departments\u0000or academic studies normally use physics-based numerical models to predict S2S\u0000sea surface temperature and corresponding climate indices, such as El\u0000Ni~no-Southern Oscillation. However, these models are hampered by\u0000computational inefficiencies, limited retention of ocean-atmosphere initial\u0000conditions, and significant uncertainty and biases. Here, we introduce a novel\u0000three-dimensional deep learning neural network to model the nonlinear and\u0000complex coupled atmosphere-ocean weather systems. This model incorporates\u0000climatic and temporal features and employs a self-attention mechanism to\u0000enhance the prediction of global S2S sea surface temperature pattern. Compared\u0000to the physics-based models, it shows significant computational efficiency and\u0000predictive capability, improving one to three months sea surface temperature\u0000predictive skill by 13.7% to 77.1% in seven ocean regions with dominant\u0000influence on S2S variability over land. This achievement underscores the\u0000significant potential of deep learning for largely improving forecasting skills\u0000at the S2S scale over land.","PeriodicalId":501166,"journal":{"name":"arXiv - PHYS - Atmospheric and Oceanic Physics","volume":"18 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142215374","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：