Ruo-Qian Wang

Turbulent water flowing over melting ice carves scallops that migrate downstream. A flume experiment has recorded one such field growing from flat ice to a self-similar state, but reproducing it numerically is challenging: wall-resolved simulation predicts the wall heat flux but is too expensive to resolve the morphology, while steady Reynolds-averaged Navier–Stokes (RANS) computation is affordable but wrongly places the melt maximum near the crest and fails to reproduce the scallops. We develop a hybrid RANS–Stefan scheme in which steady RANS computes the flow and advances the ice interface in two dimensions, and wall-resolved large-eddy simulation (LES) on a few frozen scallop shapes supplies a correcting closure. From flat seeded ice, the model can reach the laboratory equilibrium and develop a wavelength that reproduces the laboratory’s observation. The reproduced self-similar interface waves are shown to be stable. The flow field is also favorably compared against lab experiments, with the melt maximum correctly positioned downstream of the trough. Fourteen runs over U = 0.30–1.20 m s-1 and ΔT = 0.3–2.0 K were performed to reveal the conditions under which the self-similar state appears and stabilizes: it is maintained at small temperature difference and drains into long waves under strong driving.

Calibrating expensive engineering simulators requires tuning model parameters to match reference data, a process heavily constrained by computational costs. This study investigates whether Large Language Models (LLMs), conditioned on physical diagnostic descriptions of simulation errors, can propose informative parameter updates more efficiently than conventional derivative-free optimizers. We test this on a cylinder-wake Large-Eddy Simulation and a damped-oscillator inverse problem. Physics-labeled prompts are evaluated against four text ablations (removed, anonymized, padded, misdirected) and standard optimizers (Nelder–Mead, Bayesian optimization, CMA-ES, and random search). Results demonstrate a strong early semantic advantage: the first physics-informed LLM proposal improves the initial design in 23 of 25 trials. After five evaluations, the median calibration error is 21.9 times lower than physics-free controls. However, this advantage is transient—disappearing by fifteen evaluations—and inherent LLM decoder variability yields a 9.8-fold performance spread. To address this, we introduce a hybrid workflow that uses the LLM as a warm-start, passing accumulated evaluations to a conventional optimizer. This two-stage numerical continuation policy reaches target accuracy in just 9 simulator runs, outperforming the LLM alone (12 runs), Nelder–Mead (17 runs), and Bayesian optimization (>60 runs). Supported by sensitivity analysis validating the test cases, this hybrid physics-labeled LLM scheme proves highly effective as a calibration agent under strict evaluation budgets.

Passive microwave radiances near 23.8 GHz constrain lower‐tropospheric water vapor, but this protected window lies adjacent to fifth‐generation (5G) telecommunications spectrum. We quantify how projected terrestrial 5G leakage propagates through county‐scale deployment projections, directional antenna geometry, synthetic contamination of AMSU‐A channel 1 radiances, and UFS–JEDI assimilation. In a representative 2040 scenario, radio‐frequency interference (RFI) over metropolitan corridors produces brightness‐temperature perturbations up to 3.75 K. For Hurricane Ida's extratropical remnants, low‐efficiency 2040 RFI altered the 12 hr forecast relative to a clean control, with local precipitation changes up to about 10 mm and 2 m temperature differences exceeding C. Forecast changes appeared near interference sources and downstream, including regions without mapped 5G infrastructure. The results identify two pathways by which adjacent‐band emissions can affect high‐impact forecasts: biased radiances retained by assimilation and RFI‐triggered quality‐control rejection. Forecast sensitivity depends on emissions, network efficiency, and RFI‐aware assimilation.

Passive microwave observations near 23.8 GHz are central to numerical weather prediction but lie adjacent to frequencies increasingly used for fifth-generation (5G) telecommunications. Here we present the first end-to-end quantification of how terrestrial 5G interference affects weather forecasts, particularly for extreme weather events, using an operational forecasting system. By assimilating synthetically contaminated microwave radiances into a data assimilation framework, we show that aggregate out-of-band emissions introduce structured biases that propagate dynamically through the forecasting system. In simulations of Hurricane Ida’s extratropical transition, these biases trigger the rejection of moisture-sensitive observations and distort atmospheric states, producing precipitation errors of up to 15% and near-surface temperature deviations exceeding 2 °C, even far from interference sources. Our results demonstrate that anthropogenic spectral interference can systematically degrade forecasts of extreme weather, revealing a previously unquantified vulnerability in global early warning systems.

Advancements in Intelligent Transportation Systems (ITS) have led to innovative solutions for planning optimization, efficiency enhancement, and resource allocation in transportation networks, which are demonstrated in applications such as smart parking lot management and electric vehicle (EV) charging station allocation, where improved decision-making and system-wide optimization have been achieved. However, as these systems evolve, the demand for better adaptability and coordination continues to grow to maximize their overall effectiveness and efficiency. To achieve this, we propose the Multi-Personality Multi-Agent Meta-Reinforcement Learning (MPMA-MRL) framework. This approach incorporates multiple meta-trained, meta-tested explainable personality policies, which are deployed to each agent. A personality selector is trained and deployed on each agent to optimize the overall performance. MPMA-MRL is superior than traditional methods in terms of the adaptability and coordination in ITS by leveraging improved information from the environment, more practical coordination among agents, faster adaptation speed to intermediate tasks, and more appropriate allocation and planning. The proposed framework is evaluated in the applications of parking lot optimization and EV charging station allocation. Its broader impact on multi-agent smart systems is analyzed to demonstrate its generalizability. The results demonstrate that in parking lot optimization, MPMA-MRL significantly reduces the time required to direct all vehicles to available parking spots. In EV charging station allocation, MPMA-MRL effectively minimizes waiting times at charging stations. Moreover, in both applications, MPMA-MRL exhibits enhanced adaptability to previously unseen scenarios, improving its applicability.

The intensification of climate change poses significant threats to coastal regions worldwide, manifesting in increased storm frequency, sea level rise, and consequent flooding risks. This study addresses the urgent need for innovative monitoring strategies by introducing an advanced coastal hazard monitoring system specifically designed for areas with underdeveloped monitoring infrastructure. Employing a blend of traditional methods and cutting-edge technologies, including the Segment Anything Model (SAM) for high-resolution image segmentation and Dynamic Mode Decomposition (DMD) for pattern recognition, we provide a comprehensive assessment of coastal water dynamics. The study highlights the application of SAM in identifying water-land boundary despite challenges such as image distortion and variable lighting conditions. Additionally, the innovative use of monoplotting with DEM provides a robust framework for accurate mapping in complex coastal terrains. This research advances our understanding of coastal dynamics under the impact of climatic changes and sets a new benchmark for environmental monitoring, offering substantial improvements over traditional methodologies by integrating technological advancements with practical fieldwork. The findings demonstrate significant implications for disaster preparedness and the sustainable management of coastal regions, emphasizing the necessity of adopting advanced technologies to enhance the resilience of vulnerable coastal communities against the escalating threats posed by climate change.