Early meteorological in-situ observation is obtained by manual timing recording, and the temporal resolution is low. The above leads to limitations in the quality of historical data produced by land-surface data assimilation systems. Aiming at the problem of low temporal resolution of in-situ observation in the past, we propose a deep learning-based method to improve the resolution of observation time series, Deep Interpolation. DI introduces a training strategy using uncertainty to weight loss functions to solve the "seesaw phenomenon" in multi-task learning and realize collaborative temporal downscaling of multiple observation variables, including temperature, surface pressure, relative humidity, and wind speed. The experimental results in China show that the DI has achieved significant performance improvement compared with the linear interpolation method and ERA5-land, and the average correlation coefficient is as high as 0.934. DI can accurately capture the hourly fluctuation characteristics of various meteorological variables, especially in high latitude, grassland, and cultivated land regions, showing more vital correction ability and more stable model performance. This study provides an effective solution for improving the temporal resolution of historical meteorological observation data. It lays a foundation for further enhancing the accuracy and reliability of climate models and land surface assimilation systems. Improving the temporal resolution of meteorological observation data can better support the research on climate change, the construction of disaster early warning systems, and decision-making in many fields, such as agriculture and energy.

Previous tornado climatology studies often rely on tornado reports, which can suffer from underreporting bias due to heavy reliance on human reporting. This study aims to develop a new tornado climatology based on storm object data built from the Multi-Radar/Multi-Sensor reanalysis and Rapid Refresh modelsimulated near storm environment (NSE) variables. The dataset includes millions of storm clusters from 2005 to 2011, each characterized by over 1,000 radar and NSE variables and derivatives, making it ideal for developing machine learning models to classify tornadic storms. By using storm cluster data collocated with tornado reports, ML models are built using two methods: Random Forest (RF) and TabNet, a deep learning method. Three sub-models for each method, with different tornado damage-rating predictions are developed to understand the effects of sample size and model setup on performance. Both methods produce similar results, with overall Critical Success Index scores ranging from 0.720 to 0.826. TabNet predictions emphasize radar-related predictors, whereas RF predictions rely more on environmental variables with physical relationships to rotating thunderstorms. The RF model slightly outperforms TabNet and is therefore used to assign tornado probabilities to all remaining clusters. The new tornado climatology based on this RF model aligns well with observed data in terms of the magnitude of weak and strong tornadoes occurring each year. The spatial distribution of tornado days also resembles observations, providing confidence in the model's use for future applications.

A document by Md MIRAZ HOSSAIN. Click on the document to view its contents.

Precursory spurious arrivals, commonly observed in ambient noise correlations, are generated by near-field noise sources. We have developed an inversion method to evaluate the noise source distribution based on the precursory waves. This method is applied to the BASIN experiment in Los Angeles, revealing that the noise sources show coherent patterns with features like faults and structure boundaries. Our spectral analysis indicates that the energy of the noise source in generating the precursory signal is predominant at higher frequencies, suggesting a shallow source(¡200m). We conclude that near-field noise is primarily produced by scattering from geological structures with significant velocity contrasts, such as faults at shallow depths. This method offers a new way to map faults using ambient noise correlations.

Knowledge of glacier ice volumes is crucial for constraining future sea level potential, evaluating freshwater resources, and assessing impacts on societies, from regional to global. Motivated by the disparity in existing ice volume estimates, we present IceBoost, a global Machine Learning framework to model individual glacier ice thickness distributions. IceBoost is a gradient-boosted tree trained with 3.7 million global ice thickness measurements and an array of 34 numerical features. The model’s error aligns within 10% of existing models outside polar regions and is up to 30-40% lower at high latitudes. We find that providing supervision by exposing the model to available glacier thickness measurements reduces the error by up to a factor 2 to 3. A feature ranking analysis reveals that geodetic information are the most informative variables, while ice velocity can improve the model performance by 6% at high latitudes. A major feature of IceBoost is its ability to generalize beyond the training domain, producing meaningful ice thickness distributions across all global regions, including ice sheet peripheries.

Understanding the physical processes driving deep convective storms is an essential prerequisite for understanding the wider tropical atmosphere. Cold pools driven by rainfall evaporation are a ubiquitous feature of Mesoscale Convective Systems (MCSs) that have especially pronounced upscale effects in the climate of the West African Sahel, through their modulation of the regional monsoon circulation. The role of cold pools in determining the dynamics, lifecycle and propagation of tropical MCSs themselves, however, remains debated. Here we probe the feedback of cold pools on Sahelian MCS characteristics through a 40 day, 2.2 km convection-permitting Met Office Unified Model sensitivity experiment in which rainfall evaporation is switched off in the microphysics scheme. Storms generated in the sensitivity experiment subsequently show strongly suppressed cold pool density currents versus a Control simulation. Yet we find no statistically significant difference between the diurnal cycles of MCS counts, with continued nocturnal propagation of storms in the experiment, and a reduction of MCS speeds by 1.7 ms-1 caused by a similar slow-down in the African Easterly Jet due to changes to the large-scale meridional temperature gradient. Changes to composite updraft geometry are consistent with the role of cold pool horizontal vorticity in balancing the strong low-level wind shear characteristic of the region. Remarkably though, we find no sensitivity of the positive scaling of MCS rainfall with shear to cold pools, with reduced entrainment-dilution under stronger shear conditions remaining the fundamental physics driving the relationship. Our results help to disentangle processes in the Sahel in which cold pools play a driving role (nocturnal rainfall intensification, regional circulation) from those in which they are passive actors, which we find primarily to be MCS development, propagation, and rainfall-shear scalings.

Leigh Stadlera, Songo Anele Benyaa, Gina Ziervogela, Petra Holdena

C. Yang1,2 and J. Xu31CAMS & China Re CRM · Joint Open Lab on Meteorological Risk and Insurance, Beijing, China.2Faculty of Geographical Science, Beijing Normal University, Beijing, China.3Qingdao Joint Institute of Marine Meteorology, Chinese Academy of Meteorological Sciences, Beijing, China.Corresponding author: C. Yang ([email protected]), J. Xu ([email protected])Key Points:A modified Rankine vortex model with four free parameters is adopted to reconstruct wind fields along the tropical cyclone tracksA Bayesian hierarchical model with a latent neural network is developed to estimate the parameters based on the best-track data setBayesian model averaging is used to predict wind field parameters for synthetic tropical cyclones over the western North Pacific basinAbstractTropical cyclones (TCs) are one of the biggest threats to life and property around the world. Accurate estimation of TC wind hazard requires estimation of catastrophic TCs having a very long return period spanning up to thousands of years. Since reliable TC data are available only for recently decades, stochastic modeling and simulation turned out to be an effective approach to achieve more stable hazard estimates. In common practice, hundreds of thousands of synthetic TCs are generated first, then wind fields are reconstructed along synthetic TC tracks for hazard estimation. A Bayesian hierarchical modeling approach to the reconstruction of TC wind field is proposed. A modified Rankine vortex is adopted as the wind field model, of which the four free parameters are modeled simultaneously through a multi-output neural network as a latent process of the wind field. The four parameters are finally represented, spatially and temporally, by a set of neural network weights, The Bayesian model averaging technique is used for parameter estimation and wind field reconstruction, based on a ensemble of maximum a posteriori estimates of the set of weights. Together with previously proposed algorithm for synthetic TC simulation, a two-stage scheme for TC wind hazard estimation has been formed, which is based on best-track data only and thus is highly consistent. Application of this scheme to the offshore waters in the western North Pacific basin shows inspiring performance and great flexibility for various purposes of TC wind hazard estimation.

Spaceborne lidar offers unique advantages for improving global estimates of fine particulate matter PM2.5, traditionally limited by critical data gaps in the vertical dimension. Here, we present a new method to retrieve PM2.5 relying on ensembles on aerosol extinction available within the GEOS Aerosol Data Assimilation. This study uses 1064-nm backscatter lidar data from the NASA Cloud-Aerosol Transport System (CATS) and model priors from the GEOS model. First, we developed a 1-D ensemble-based variational technique (1-D EnsVar) to perform vertically-resolved retrievals of speciated aerosol extinction and surface PM2.5. Next, we evaluated the performance of 1-D EnsVar retrievals of PM2.5 and extinction through an independent validation using measurements from spaceborne, airborne, and ground-based platforms. This approach overcomes traditional limitations by leveraging the strengths of complementary vertical aerosol information from CATS and GEOS to better resolve speciated aerosol optical properties and mass. With the advantage of active remote sensing, this approach is fully capable of performing aerosol retrievals during both daytime and nighttime scenes. Given the unique capability of CATS to process vertical profile data in near real-time, this work demonstrates the powerful utility of spaceborne lidar for improving air quality forecasting. While this pilot study is not yet performed within a cycling data assimilation system, the algorithm developed here can easily be integrated in such systems. Results presented may be useful in other applications as well, including the validation of aerosol transport models, improving passive satellite retrievals of PM2.5, and developing data assimilation techniques for future lidar platforms.

Planetary analog simulations are a powerful exercise for understanding the utility of deployable instruments, their operational protocols, and the visualization of data products during ExtraVehicular Activities (EVAs). This paper presents results of a field campaign by the Remote, In Situ and Synchrotron Studies for Science and Exploration-2 (RISE2) team to Kilbourne Hole, New Mexico in March/April 2023 to test the utility of a portable thermal infrared (TIR) hyperspectral imager (HSI) during four EVA simulations. The HSI provides emitted radiance spectra from 7 to 14 µm to map composition, which aids in sample selection. Four pairs of analog astronauts performed a mock EVA at three stations with field deployable instruments including the HSI. The HSI was found to be a useful tool for performing reconnaissance observations, field site documentation, and sample selection for visibly indistinct materials. From these analog simulations we prioritize two recommendations for use of HSI in crewed missions. First, HSI-derived data products should be tailored for the specific science objectives and/or sampling objectives of the mission to expedite interpretation and decision-making. Second, pre-EVA reconnaissance using HSI data collected from a remotely operated rover could further enhance the usability of HSI products.

Bifurcations in river-dominated deltas are the main actors in the routing of water and sediments throughout the fluvial network. In spite of previous acknowledgments of their importance, we still lack a comprehensive framework on how bifurcation geometry affects the water and sediment partitioning. To investigate this issue, we firstly combine previously calibrated 2D hydrodynamic simulations on the Wax Lake Delta with a Lagrangian particle-tracking model, quantifying the partitioning of water and sediments with different buoyancy at five bifurcations and their correlations with differences in channel width, branching angle and inlet bed elevation between the downstream branches. We compare the sediment partitioning at bifurcations with available field data to validate our methodology. We then employ the same modeling tools on a simplified geometry, whose geometrical and hydraulic features resemble those of the bifurcations in the Wax Lake Delta. Model results show that the branching angle does not affect the partitioning of water and sediments. The combined effect of asymmetries in the channel width and inlet bed elevation is captured by a simple linear formula that accurately estimates the partitioning of water at bifurcations returned by the 2D calibrated hydrodynamic simulations. Our results also highlight the key role played by transverse gradients in the bathymetry of the upstream channel in determining the partitioning of sediments, suggesting that deeper portions of the cross-section of the upstream channel can cause a proportionately larger fraction of sediments with larger Rouse number to be routed towards the corresponding bifurcate.

Predicting the occurrence of coherent blocking structures in synoptic weather systems remains a challenging problem that has taxed the numerical weather prediction community for decades. The underlying factor behind this difficulty is the so-called "loss of hyperbolicity" known to be linked with the alignment of dynamical vectors characterizing the growth and decay of flow instabilities. We introduce measures that utilize the close link between hyperbolicity, the alignment of Lyapunov vectors, and their associated growth and decay rates to characterize the dynamics of persistent synoptic events in the mid-troposphere of the Southern Hemisphere. These measures reveal a general loss of hyperbolicity that typically occurs during onset and decay of a given event, and a gain of hyperbolicity during the persistent mature phase. Facilitating this analysis in a high-dimensional system first requires the extraction of the relevant observed coherent structures, and the generation of a reduced-order model for constructing the tangent space necessary for dynamical analysis. We achieve this through the combination of principal component analysis and a non-parametric, temporally regularized, vector auto-regressive clustering method. Analysis of the primary blocking sectors reveals hyperbolic dynamics that are consistent between metastable states and whose dynamics span the tangent subspace defined by the leading physical modes. We show that these diverse synoptic features are manifest via common spatially dependent attractors as determined by tangent space dynamics. Our results are not only important for dynamical approaches applicable to high-dimensional multi-scale systems, but are also relevant for the development of modern operational ensemble numerical weather prediction systems.

The Great Lakes of North America form one of the largest freshwater systems on Earth, exhibiting seasonal water level fluctuations that exceed 0.5 meters. These fluctuations pose substantial challenges for coastal resilience, flood risk management, and navigation planning. Accurate seasonal forecasting of lake levels using traditional mechanistic models is challenging due to the complex physical mechanisms and coupled hydroclimatic processes involved. Recently, deep learning has gained prominence in geoscience applications for its ability to recognize intricate patterns within multiphysical datasets. Here, we introduce a novel Dual-Transformer deep learning framework, tested on Lake Superior—the largest of the five Great Lakes. This architecture integrates two modified Transformer models: the Prophet, which predicts underlying trends, and the Critic, which refines the Prophet’s predictions. The final lake level prediction is derived by weighting the outputs of both models through a Multi-Layer Perceptron (MLP), jointly trained with the Prophet and Critic to enhance overall accuracy. Our results demonstrate that the Dual-Transformer model, which uses seven atmospheric and lake features, achieves unprecedented accuracy in seasonal forecasting in the testing dataset, attaining a correlation coefficient of 0.97 and a root mean square error (RMSE) of 4 cm for forecasts up to six months ahead. Additionally, the Dual-Transformer model runs six orders of magnitude faster than conventional mechanistic models, producing results in less than one second on a typical personal computer. These findings suggest our deep learning framework provides an efficient and reliable tool for real-time lake level forecasting, with significant implications for water management and disaster mitigation.

Diapycnal mixing in the ocean interior has diverse numerical representations in extant global ocean models. These representations affect the simulated transport and storage of oceanic tracers in ways that remain little studied. Here we present the impacts of three different tidal mixing schemes in thousand-year-long simulations with the NEMO global ocean model at one-degree resolution. The first scheme includes local bottom-intensified mixing at internal tide generation sites and a constant background diffusivity. The second explicitly includes both local and remote tidal mixing, with no background diffusivity. The third scheme is identical to the second but has the added contribution of bottom-trapped (subinertial) internal tides, known to be important in polar regions. The three simulations show broadly similar circulation and stratification but important regional differences. Explicit representation of remote tidal mixing strengthens the Atlantic Meridional Overturning Circulation by up to 1.5 x 106 m3 s-1. Inclusion of bottom-trapped internal tides reduces heat reaching Antarctica by eroding Circumpolar Deep Water at southern high latitudes, and reduces the mean age of the global deep (> 2 km) ocean by 10%. The results call for more observational constraints on polar ocean mixing, and point to multi-faceted climatic repercussions of tidal mixing representations.

Pacific decadal variability (PDV), low-frequency changes in Pacific sea surface temperatures (SSTs), significantly impacts global climate. However, disentangling anthropogenic effects upon PDV is challenging since both vary on similar time scales. Using single-forcing climate model large ensembles, we find that anthropogenic forcing primarily drives a spatially-varying pattern of mean-state change in North Pacific SST that project onto leading PDV patterns, principally the Pacific Decadal Oscillation (PDO) and North Pacific Gyre Oscillation (NPGO). In fact, when the trend is determined by the model ensemble mean, there is no forced change of the PDV modes. However, analysis of single model realizations, where the mean-state trend cannot be cleanly identified, suggests an apparent anthropogenic change in NPGO decadal variability. This suggests that observed PDV responses to anthropogenic forcing may be erroneously convolved with the background trend pattern. Therefore, correctly determining the mean-state trend is a necessary precursor for identifying forced changes to PDV.

Despite the importance of the Southern Ocean carbon sink, its response to future atmospheric CO2 perturbations and warming remains highly uncertain. In this study, we use six state-of-the-art Earth system models to assess the response of Southern Ocean air-sea CO2 fluxes (FCO2) to a rapid atmospheric forcing increase and subsequent negative emissions in an idealized carbon dioxide removal reversibility experiment. We find that during positive emissions, the region north of the Polar Front only takes up atmospheric CO2 for 30-50 years before reaching equilibrium; surface stratification and reduction of CO2 solubility with warming diminishes ocean CO2 uptake in this region. In contrast, south of the Polar Front, the upper ocean continues to take up CO2 until the end of positive emissions at 140 years. Sea-ice loss and the accumulation of anthropogenic dissolved inorganic carbon in the upper ocean reduce the upwelling-driven seasonal CO2 outgassing, leading to a stronger Antarctic CO2 sink. CO2 removal triggers a CO2 uptake reduction that slowly converts the Southern Ocean into a CO2 source which persists for at least 50 years post-mitigation. Furthermore, we find that model sensitivity to atmospheric perturbation is closely linked to seasonal FCO2 dynamics. Specifically, models with a thermally dominated pCO2 seasonal cycle exhibit nearly twice the sensitivity to atmospheric perturbations compared to non-thermal models. Our findings further emphasize the necessity of accurate model representation of the seasonal CO2 dynamics for appropriately simulating the future Southern Ocean carbon sink.

During the Asian Summer Monsoon Chemical and Climate Impact Project (ACCLIP), National Aeronautics and Space Administration (NASA) WB-57 high altitude research aircraft observed strong turbulence at an altitude of 15 to 18 km near the outflow of the Super Typhoon Hinnamnor in the Northwest Pacific. Our analyses based on the trial version of the “RRJ-ClimCORE” mesoscale reanalysis have revealed that a thin layer of strong vertical wind shear (VWS) extending just below the tropopause along the upper surface of the outflow from the typhoon eyewall provides favorable conditions for shear instability. Further analyses of the mesoscale background conditions based on RRJ-ClimCORE as well as the geostationary satellite Himawari-8 cloud imagery suggest that the convolution of such small-scale processes as convectively generated concentric gravity waves and shallow convection appearing as radially-banded cirrus clouds may have further increased the potential for turbulence generation. Absolute vorticity analysis suggests the occurrence of inertial instability to the north of the typhoon, which explains the development of radially-banded cirrus clouds as well as the layer of strong VWS around the flight altitudes. The present study thus demonstrates the advantages of the hourly output of RRJ-ClimCORE over 3-hourly output operational mesoscale analysis for investigating source mechanisms of turbulence in the upper-troposphere and lower-stratosphere.

• Electron precipitation in Jupiter's northern auroral oval can increase microwave opacity at frequencies observed by the Juno MicroWave Radiometer, appearing as cold spots in brightness temperature. • Precipitation of 32 keV-10 MeV electrons results in ionization of the lower stratosphere below the methane homopause, producing short-lived hydrocarbon ions. • Rapid changes in MWR antenna temperatures over consecutive spacecraft spins highlight short-time scale temporal variations in the ionospheric medium. They imply downward electron fluxes whose fluxes vary within 10 s.