The rapid evolution of the low altitude economy, characterized by unmanned aerial vehicles (UAVs) and urban air mobility operations within non-controlled airspace classes such as G and W, has positioned it as a pivotal driver of modern economic growth. This sector relies heavily on safe and efficient flight operations in the lower atmospheric boundary layer, where meteorological conditions—particularly turbulence, wind shear, and microbursts—pose significant risks. The integration of meteorology with communication, navigation, and surveillance systems forms the foundational support for the sustainable development of the low altitude economy. However, current capabilities in high spatiotemporal resolution monitoring and forecasting of aviation hazardous weather in these airspaces remain inadequate, presenting critical challenges to safety and operational efficiency. This article systematically reviews global research progress in low altitude economy meteorology, analyzes key scientific and technological hurdles, and explores frontier issues to guide future innovations. Emphasis is placed on the interplay between atmospheric boundary layer processes, advanced sensing technologies, artificial intelligence (AI)-driven forecasting, and computational fluid dynamics (CFD) simulations, all aimed at enhancing the safety and scalability of low altitude economy activities.
The atmospheric boundary layer (ABL), extending up to approximately 1–3 km in altitude, is the primary domain for low altitude economy operations. Turbulent coherent structures within the ABL, such as convective rolls and cells, are governed by the balance between buoyancy and wind shear, quantified by the stability parameter $-z_i/L$, where $z_i$ is the boundary layer height and $L$ is the Obukhov length. For instance, in convective boundary layers (CBLs), the transition from roll-like to cellular structures occurs around $-z_i/L \approx 25$, as observed in large eddy simulations (LES) and field experiments. The energy cascade in turbulent flows, described by the Kolmogorov theory, underpins the dissipation rate $\epsilon$, which is critical for assessing turbulence intensity affecting UAVs: $$\epsilon = \frac{u’^3}{l},$$ where $u’$ is the turbulent velocity fluctuation and $l$ is the integral length scale. Advances in remote sensing, such as lidar and radar, have enabled real-time monitoring of these structures, yet gaps persist in predicting their evolution under non-stationary conditions, such as during sunrise or sunset transitions, which are common in low altitude economy scenarios.
Research on clear-sky boundary layers has leveraged high-resolution radiosonde data and LES to characterize the diurnal variation of boundary layer height $z_i$. For example, studies using normalized signal-to-noise ratio profiles from wind profiler radars have derived minute-scale $z_i$ estimates, revealing its correlation with pollutant dispersion—a key factor for UAV operations in urban areas. The relationship between $z_i$ and surface particulate matter (PM) concentration can be modeled as: $$[PM] = k \cdot z_i^{-\alpha},$$ where $k$ and $\alpha$ are empirical constants. This has implications for visibility and UAV performance in the low altitude economy. However, the lack of continuous, high-altitude turbulence observations remains a limitation, driving the need for integrated sensing networks.
In terms of hazardous weather monitoring, low-altitude turbulence—categorized as clear-air turbulence (CAT) and convection-induced turbulence (CIT)—has been a focal point. Traditional methods like pilot reports (PIREPs) are being supplemented with AI algorithms. For instance, random forest models have achieved high accuracy in CAT prediction by integrating multi-source data, including radar reflectivity $Z$ and eddy dissipation rate (EDR): $$\text{EDR} = \left( \frac{\epsilon}{\nu} \right)^{1/4},$$ where $\nu$ is the kinematic viscosity. Similarly, microburst detection has evolved from Doppler radar-based systems, such as the Terminal Doppler Weather Radar (TDWR), to AI-enhanced forecasting models that use Bayesian approaches to reduce false alarms. The following table summarizes key advancements in low-altitude hazardous weather monitoring:
| Hazard Type | Monitoring Technology | Key Parameters | Advances |
|---|---|---|---|
| Low-Altitude Turbulence | Lidar, Radar, AI Models | EDR, $u’$, $l$ | Machine learning integration for real-time forecasting |
| Wind Shear | Doppler Radar, UAV Sensors | $\Delta V/\Delta z$, $\Delta \theta/\Delta z$ | Automated detection algorithms and risk assessment |
| Microbursts | TDWR, LES Simulations | Downdraft velocity $w_d$, outflow radius $r$ | Physical-informed neural networks for prediction |
Field experiments have played a crucial role in validating these technologies. For example, campaigns like the Joint Airport Weather Studies (JAWS) and urban wind shear observations using lidar have provided datasets for model calibration. In one study, lidar-based wind shear algorithms reduced false alarms by 30% in airport approaches, enhancing safety for low altitude economy operations. The equation for wind shear magnitude $S$ is: $$S = \sqrt{\left( \frac{\partial u}{\partial z} \right)^2 + \left( \frac{\partial v}{\partial z} \right)^2},$$ where $u$ and $v$ are horizontal wind components. These efforts underscore the importance of multi-platform observations, including UAV-mounted sensors, which offer agile data collection in complex terrains.
Large eddy simulation (LES) has emerged as a powerful tool for resolving turbulent structures in the ABL. The filtered Navier-Stokes equations in LES are: $$\frac{\partial \tilde{u}_i}{\partial t} + \tilde{u}_j \frac{\partial \tilde{u}_i}{\partial x_j} = -\frac{1}{\rho} \frac{\partial \tilde{p}}{\partial x_i} + \nu \frac{\partial^2 \tilde{u}_i}{\partial x_j^2} – \frac{\partial \tau_{ij}}{\partial x_j},$$ where $\tilde{u}_i$ is the filtered velocity, $\tilde{p}$ is the pressure, and $\tau_{ij}$ is the subgrid-scale stress tensor. GPU-accelerated LES models, such as FastEddy, have reduced computational costs by two orders of magnitude, enabling high-resolution simulations of urban airflow at 10 m scales. This is vital for the low altitude economy, as it allows for precise modeling of building-induced turbulence and wind patterns that affect UAV trajectories. Similarly, urban micro-scale CFD simulations integrate LES with Reynolds-averaged Navier-Stokes (RANS) approaches to predict local wind fields: $$\frac{\partial U_i}{\partial t} + U_j \frac{\partial U_i}{\partial x_j} = -\frac{1}{\rho} \frac{\partial P}{\partial x_i} + \frac{\partial}{\partial x_j} \left( \nu_t \frac{\partial U_i}{\partial x_j} \right),$$ where $U_i$ is the mean velocity and $\nu_t$ is the turbulent viscosity. These models have been coupled with mesoscale weather predictions to provide dynamic boundary conditions, improving forecast accuracy for low-altitude flights.
Path planning for UAVs in the low altitude economy has evolved to incorporate meteorological data. Optimization algorithms, such as genetic algorithms or rapid-exploring random trees (RRT), minimize a cost function $C$ that includes energy consumption $E$, time $T$, and risk $R$: $$C = \alpha E + \beta T + \gamma R,$$ where $\alpha$, $\beta$, and $\gamma$ are weighting factors. CFD-derived wind fields inform risk assessments, defining no-fly zones based on turbulence intensity $I$: $$I = \frac{\sigma_u}{U},$$ where $\sigma_u$ is the standard deviation of wind speed and $U$ is the mean wind speed. However, real-time path planning remains challenging due to computational delays in CFD simulations, prompting the use of machine learning for rapid wind field predictions.
Looking ahead, several frontier scientific issues demand attention to advance the low altitude economy. First, the theory of near-neutral boundary layer turbulence requires refinement to address non-stationary conditions, such as typhoon boundaries or diurnal transitions. The momentum flux $\tau$ in these layers can be expressed as: $$\tau = \rho u_*^2,$$ where $u_*$ is the friction velocity, but current models struggle with baroclinic effects. Second, wet boundary layer processes, involving cloud formation and precipitation, introduce complexities due to latent heat release and droplet drag. The buoyancy term in the turbulence kinetic energy (TKE) equation becomes: $$B = -\frac{g}{\theta_0} \overline{w’\theta_v’},$$ where $\theta_v$ is the virtual potential temperature and $w’$ is the vertical velocity fluctuation. Integrated observing systems, combining cloud radar and UAV-based sensors, are needed to capture these phenomena.
Third, the development of intelligent meteorological sensing equipment is crucial for the low altitude economy. Miniaturized lidars and UAV-mounted sensors can provide high-resolution data on wind shear and turbulence, but require standardization and networking. Fourth, AI-driven short-term warning systems must evolve to handle秒级 updates. Physics-informed neural networks (PINNs) that embed the Navier-Stokes equations offer promise: $$\mathcal{L} = \mathcal{L}_{\text{data}} + \lambda \mathcal{L}_{\text{physics}},$$ where $\mathcal{L}$ is the loss function and $\lambda$ is a weighting parameter. Such models can predict microbursts with lead times of minutes, enhancing safety for low altitude economy operations.
Fifth, high-fidelity LES models operable on GPU architectures are essential for urban-scale forecasting. The Smagorinsky subgrid-scale model, with a dynamic coefficient $C_s$, improves accuracy: $$\tau_{ij} = -2 \rho (C_s \Delta)^2 |\tilde{S}| \tilde{S}_{ij},$$ where $\Delta$ is the filter width and $\tilde{S}_{ij}$ is the strain rate tensor. Coupling these with CFD enables seamless downscaling from mesoscale to street-level, critical for the low altitude economy. Sixth, intelligent CFD techniques that blend machine learning with physical models can accelerate simulations. For example, generative adversarial networks (GANs) can learn from high-fidelity CFD data to produce real-time wind fields, reducing computational costs by 80% while maintaining accuracy.
Seventh, synergistic optimization of low-altitude meteorological conditions and UAV path planning is vital. Dynamic programming approaches that adjust trajectories based on real-time wind forecasts can minimize energy use $E$: $$E = \int P(v, w) \, dt,$$ where $P$ is the power consumption as a function of airspeed $v$ and wind $w$. Finally, establishing standards for low altitude economy meteorology, such as safe distances from clouds and turbulence thresholds, will require extensive experimental data and international collaboration.
In conclusion, the low altitude economy stands to benefit immensely from advancements in meteorology, but significant challenges remain. Bridging gaps in turbulent theory, enhancing sensing capabilities, and leveraging AI and high-performance computing will be key to ensuring safe and efficient operations. Future research should focus on integrated systems that combine real-time data assimilation, physical modeling, and autonomous decision-making to support the growing demands of the low altitude economy. By addressing these frontier issues, stakeholders can unlock the full potential of low-altitude airspace, fostering economic growth while prioritizing safety and sustainability.