Wind Speed Area Research Articles

Observed Over-the-Horizon Propagation Characteristics and Evaporation Duct Inversion During the Entire Process of Tropical Cyclone Mulan (202207)

An evaporation duct, which is often occurs in the air–sea boundary layer, have a considerable impact on the remote sensing detection capability of electromagnetic (EM) systems. Tropical cyclones (TCs) substantially increase the transfer of water vapor and heat in the air–sea boundary and inevitably affect EM propagation environments. However, observations of duct environments and their effects on EM propagation during a TC process have seldom been reported. In this paper, X-band EM propagation systems that can operate continuously in the TC environment were specially developed, and observations of multiple over-the-horizon propagation links during TC Mulan (202207) were presented. The path loss (PL) of five links, and the surface variations both in Mulan’s area of high wind speed (WS) and eye were recorded. The average PL of a 53-km link was approximately 155 dB in the high WS area, while a 30 dB increase in PL was detected in Mulan’s eye. The inversion and simulation results show that the evaporation duct height (EDH) was high in the high WS area, which led to low PL in EM propagation; conversely, EDH was low inside Mulan’s eye, resulting in high PL.

Spatial heterogeneity of driving factors-induced impacts for global long-term surface urban heat island

ABSTRACT A series of empirical analytical tools have been adopted to investigate the driving mechanisms of surface urban heat islands (SUHI) on a global scale, among which spatial heterogeneity is yet to be fully elucidated. In this study, we investigated the spatial non-stationarity of the driving factors concerning surface properties, climate conditions, and urbanization processes for global long-term SUHI. First, the potential impact on SUHI was explored using global ordinary least squares regression. Geographically weighted regression (GWR) and multi-scale GWR (MGWR) from local perspectives were employed for comparison. The results show that the MGWR has the highest goodness of fit at 0.87, 0.73, 0.90, 0.74, 0.85, and 0.76 for annual day/night (AD/AN), summer day/night (SD/SN), and winter day/night (WD/WN) scales, respectively. Although both global and local schemes exhibit similar influencing magnitudes and signs on the SUHI, the MGWR is better at capturing spatial non-stationarity. Globally, for AD, AN, SD, SN, WD, and WN, the coefficients of the urban-rural vegetation index difference (ΔEVI) and surface albedo difference (ΔWSA), urban mean precipitation (MAP), wind speed (WS), population density (PD), and urban area (UA) are −0.50, +0.30, +0.16, +1.31, −0.03, and +0.03, respectively, at daytime, and −0.38, −0.33, −0.39, −0.10, +0.18, and +0.08, respectively, at night-time. Given the spatial heterogeneity of multiple factors, ΔEVI exhibits a strong mitigation effect on the SD SUHI especially in arid zones. The negative influence of ΔWSA on night-time SUHI demonstrates a strong latitudinal disparity and greater sensitivity in the equatorial zone. The positive correlations between MAP and AD/SD SUHIs have evident latitudinal and longitudinal variations. The mitigation effect of WS displayed distinct coastal amplification, especially in WD. In contrast, the PD and UA presented prominent positive impacts on night-time SUHI with less seasonal contrast.

A unified optimization control of wind farms considering wake effect for grid frequency support

This paper proposed a unified active power optimization control of wind farms under wake effect. It takes the sum of the kinetic energy variation and pitch angle variation as the optimization objective and used the particle swarm algorithm to achieve the optimization results. The main feature of the proposed method is that it unifies the kinetic energy optimization under a low wind speed area, the pitch angle and kinetic energy trade-off optimization under a medium wind speed area, and the pitch angle optimization under a high wind speed area. Combined with the de-loaded power constraint, it can flexibly reach various optimal operating states of the wind farm. The simulation results show that the proposed method optimizes the rotor speed and pitch angle in different wind speed areas, and releases kinetic energy and/or increases the output power of the wind farm to provide frequency support by switching the operating states.

Effects of Vertical Forests on Air Quality in Step-up Street Canyons

Vertical forests have been constructed as environmentally friendly solutions for air quality and urban heat problems. With an expectation to provide insights into the value of vertical forests, we examined the impact of these structures on flows and distribution of fine particles in step-up street canyons using a computational fluid dynamics (CFD) model. The CFD model was rigorously validated against wind-tunnel experiments and accurately simulated particle deposition efficiency by tree leaves under varying wind speed and leaf area density (LAD) conditions, as well as the flow structures in step-up street canyons. We analyzed the characteristics of flows and fine particle distributions in step-up street canyons for different LADs and tree deposition velocities. Despite a substantial reduction in fine particles through dry deposition, the fine particles emitted from two line-type sources within the street canyon increased by 20-25% in the canyon's lower layer due to a 14-20% reduction in wind speed. This study sheds light on the complex flow structures and pollutant distributions surrounding vertical forests and highlights the need for a comprehensive evaluation and optimal design of these structures.

Improving the self-starting and operating characteristics of vertical axis wind turbine by changing center distance in part of blades

Compared with the horizontal axis wind turbine (HAWT), the vertical axis wind turbine (VAWT) has the advantages of low noise operation and no wind yaw device. However, the poor self-starting performance of VAWT is one of key problems restricting its development. In order to improve the self-starting capability of VAWTs in urban wind areas, this paper proposed a VAWT model with multi-blade local variable radius. The main object of this study is to research the effect of reducing the partial blade radius on the self-starting performance and power coefficient of the VAWT. In this study, the partial blade radius of the VAWT models was reduced to 0.375, 0.5, and 0.625 times of the outer layer. The dynamic torques and velocity contours of the different models were researched by experiments and numerical simulations to analyze the reason why reducing the partial blade radius affects the power coefficient of the VAWT. The static torque and static pressure contours of different models were researched to compare the self-starting capability of the models. According to the results, the minimum static torque values of the three innovative VAWT models were increased by 0.1, 1.8, and 2.8 times, respectively, compared to the traditional VAWT with only the outer layer. Furthermore, the power coefficients of three models were increased for tip speed ratios (TSR) less than 2.2. Therefore, the new models can improve the self-starting capability of VAWTs, and the higher power coefficients at low TSRs make them more advantageous for applications in urban buildings and low wind speed areas.

Synoptic view of an unprecedented red Noctiluca scintillans bloom in the Beibu Gulf, China

An unprecedented red Noctiluca scintillans (RNS) bloom in February 2021 in the Beibu Gulf, China was investigated using multi-sensor imagery. The location and spatial extent of the RNS patches were delineated based on the spectral signature. A combination of satellite remote sensing and numerical model data over the region was used to detect and monitor the development and progress of the RNS bloom and figure out the possible mechanism. Multi-sensor satellite imagery demonstrates a total coverage area of >20,000 km2 for the two-phase bloom event from Feb 13 to 26, with abrupt dissipation on Feb 17 and reappearance on Feb 20. Analysis of the initialization mechanism of the bloom suggests that nutrients conveyed from the western coast of Guangdong through Qiongzhou Strait and from convective transport in the offshore area together stimulated the outbreak of the bloom. The most distinctive finding is the negative correlation between the area of RNS patches and wind speed. The abrupt disappearance of the RNS bloom deserves further study.

Design and analysis of energy efficient wind turbine blades

Inspired by humpback whale flippers, this study presents the effect of leading-edge tubercles on the wind turbine blade. Force models for lift and drag are developed on the blade which have an important role in the wind turbine performance. The study compares the CFD simulation of fluid flow over blade with leading edge tubercles and the conventional blade. Present work also covers strength and random vibration analysis for both the blades. Furthermore, different tubercle blades have been compared for lift and drag coefficients by varying tubercle amplitude. Structural analysis shows tubercled blade is suitable from structural and strength point view. In tubercled blade overall lift to drag-coefficient ratio increased up to 8.5, whereas in conventional this ratio is about 1.7 which is found to be very low. This ratio clearly indicates that in the presence of tubercles on the blade, performance improving even in lower wind speed areas. Overall tubercled blade found to be more suitable for modern wind turbines.

Relieving water stress by optimizing crop structure is a practicable approach in arid transboundary rivers of Central Asia

With the challenges of arid transboundary river water management among Central Asian countries, it is important to evaluate the water stress and propose practicable climate adaptation strategies. This study used the Syr Darya Basin (SDB) in a Central Asia data scarce region as the study area and, conducted the first quantified analysis of the spatiotemporal characteristics of water supply and requirements at the basin and irrigation district scale; identified its driving mechanisms based on systematic monitoring data concerning reservoir regulation and complex crop structure from 2000 to 2018; and a climate change scenario combined with a crop restructuring scenario was applied to evaluate water conservation effects and propose a climate adaptation strategy from 2019 to 2030. The results showed that runoff increased by 7.45 billion m3 but the water withdrawal decreased by 3.42 billion m3 after reservoir regulation in the basin during 2000–2018. Due to the unified water allocation, it was challenging to conduct and usually exceeded the plan by 0.4–8.5 %. The total crop water requirement (TCWR) mainly derived from cotton (61.5 %), winter wheat (12.1 %) and rice (13.9 %) increased by 3.15 billion m3 (+30.7 %) and the influence of crop area (380 million m3/a) was greater than that of climate (130 million m3/a). The wind speed (27.5 %) and rice area (19.2 %) also significantly contributed to the TCWR variation. The contradiction between water supply and requirement resulted in water scarcity in the basin. From 2019–2030, the water conservation effect of winter wheat (+20 %) under the restructuring scenarios reached 28.7 mm, which helped relieve water stress under the scenario of a 0.51–0.91 °C increase. Thus, priority planting of a winter wheat rotation and low water consuming crops is a practicable approach for coping with water scarcity and adapting to climate change in arid transboundary rivers of Centra Asia.

Optimum Selection of Communication Tower Structures Based on Wind Loads & lifecycle cost analysis

Communication towers are vital assets in our daily lives as they transfer signals between cell phones facilitating communication and commerce among people and businesses all around the world. Wind loads are crucial in the communication towers design since they are tall and slender. With climate change bringing more storms and higher wind speeds, it is more crucial to research the finest tower structure that withstands such conditions with the least life cycle cost. Therefore, in this paper, a comparative case study is performed between 45 m height lattice tower and monopole tower in Egypt. Two locations were considered, the first is inside the city and the second is an open terrain. In addition, two load scenarios were investigated. The comparison parameters are the behavior under critical wind loads taking into account three wind speeds which are 100 km/hr, 130 km/hr and 140 km/hr, and life cycle cost analysis. It was found that the lattice tower behaves better under critical wind loads with a maximum tilting equal to 0.4784 degrees at location 1, load 2, and a wind speed of 140 km/hr compared to 0.5806 in the case of the monopole tower. Similarly, the lattice tower behaves better at the second location as well. However, the monopole tower has less life cycle cost with a total life cycle cost of 3,201,846.80 EGP compared to 4,380,419.91EGP in case of lattice tower. Therefore, based on the location, wind speed, and available land area, and life cycle cost assessment, the optimum tower structure could be selected.

Role of source terms in parameterizing wave decay in the marginal ice zones

Wave decay in the marginal ice zones (MIZ) plays a crucial role in shaping the Arctic Ocean behaviours. Parameterizing wave decay in the MIZ is indispensable in climate models for better capturing the climate change in the Arctic. However, current wave decay parameterizations were developed without considering the influences of wave source terms. This study investigated the role and contributions of different source terms in shaping the MIZ wave decay based on sensitivity simulations. Simulation results show that the ice-induced damping term is the main contributor to MIZ wave decay. The wind input source term also plays an important role in the wave evolution in the MIZ. During high wind speeds (wind speed>10 m/s), wind input was found to increase the wave height by about 10% on average (relative to the wave height decayed by ice), irrespective of the sea ice concentration. Wind input contributes up to 30% in high wind speed areas with low sea ice concentration. Meanwhile, the contributions from wind input term to MIZ wave are more important when winds blow from the open water to MIZ. However, the angle between winds and waves is insignificant in shaping contributions from wind input. Low SIC and high wind speed conditions also favour the enhancement of wave spreading. In general, the contributions from nonlinear wave–wave interactions and dissipation source terms were found to play a minor role in shaping MIZ wave decay. However, the role of nonlinear wave–wave interactions can be important in shaping the wave spectrum. Thus, it is suggested to update the model wave decay parameterizations to accommodate the role played by wind input source term, in particular under high wind speed conditions.

Study on Accuracy of CFD Simulations of Wind Environment around High-Rise Buildings: A Comparative Study of k-ε Turbulence Models Based on Polyhedral Meshes and Wind Tunnel Experiments

It is important to create a comfortable wind environment around high-rise buildings for outdoor activities. To predict the wind environment, Computational Fluid Dynamics (CFD) has been widely used by designers and engineers. However, the simulation results of different CFD turbulence models might significantly vary. This paper researched the wind environment around a typical high-rise building and verified the accuracy of the CFD simulations based on polyhedral meshes. The differences between the simulation results of the k-ε turbulence models and those of the wind tunnel experiments were compared from the perspectives of wind speed and turbulence energy. The results show that the modified k-ε models could still not perfectly match the wind tunnel experiment results. Specifically, in the low-wind-speed areas, the simulation results of the Realizable Two-Layer K-Epsilon (RTLKE) model were the closest to the experimental results of the wind tunnels, while in the high-wind-speed areas the simulation results of the Standard Two-Layer K-Epsilon (STLKE) model were the closest to the experimental results of the wind tunnels. Therefore, it is recommended that these two k-ε turbulence models are applied under different conditions—the RTLKE model should be used to simulate low-wind areas around high-rise buildings (e.g., defining the size of the static-wind area around high-rise buildings, predicting the diffusion time of pollutants around high-rise buildings, etc.); STLKE should be used to simulate high-wind-speed areas around high-rise buildings (e.g., the high speed wind area around high-rise buildings during a typhoon, the maximum wind speed area around high-rise buildings, etc.). It is expected that findings from this research study supplement some existing high-rise building design guidance.

Species-specific efficiency in PM2.5 removal by urban trees: From leaf measurements to improved modeling estimates

The growing population in cities is causing a deterioration of air quality due to the emission of pollutants, causing serious health impacts. Trees and urban forests can contribute through the interception and removal of air pollutants such as particulate matter (PM). The dry deposition of PM by vegetation depends on air pollutant concentration, meteorological conditions, and specific leaf traits. Several studies explored the ability of different plant species to accumulate PM on leaf structures leading to the development of models to quantify the PM removal. The i-Tree Eco is the most used model to evaluate ecosystem services provided by urban trees. However, fine particulate matter (PM2.5) removal is still calculated with a poorly evaluated function of deposition velocity (which depends on wind speed and leaf area) without differentiating between tree species. Therefore, we present an improvement of the standard model calculation introducing a leaf trait index to distinguish the species effect on PM net removal. We also compared model results with measurements of deposited leaf PM by vacuum filtration. The index includes the effect of morphological and functional leaf characteristics of tree species using four parameters: leaf water storage, deposition velocity, resuspension rate and leaf washing capacity. Leaves of 11 common urban tree species were sampled in representative areas of the city of Ferrara (Italy) and at different times of the year from 2018 to 2021. This includes four deciduous broadleaf trees (Tilia cordata, Platanus acerifolia, Acer platanoides, Celtis australis), three evergreen broadleaf trees (Quercus ilex, Magnolia grandiflora, Nerium oleander), and four conifers (Thuja orientalis, Cedrus libani, Pinus pinaster, Picea abies). The results provide significant advancement in assessing PM removal using decision support tools such as models to properly select tree species for future urban tree planting programs aimed at improving air quality.

Numerical Investigation of the Power Performance of the Vertical-Axis Wind Turbine with Endplates

An H-rotor vertical axis wind turbine (VAWTs) can operate independently in any wind direction, making it aerodynamically efficient and suitable to harness wind energy in low wind speed areas. The aerodynamic efficiency of VAWTs is highly dependent on the blade geometry, especially the blade tip. Tip vortices produced at the blade tips can negatively affect the VAWT’s aerodynamic efficiency. Adding endplates to the blade tips can minimize the effects of tip vortices on VAWTs. In this paper, several endplate designs are used to evaluate the effectiveness in improving the power coefficient, Cp of a VAWT at three different tip speed ratios (TSRs) using three-dimensional computational fluid dynamics (3D CFD) simulation. The power coefficients of VAWTs with endplates are compared with the baseline model with the same geometrical parameters where the baseline VAWT model is based on the experimental model from the literature. Since the focus of this study is on the blade tip design, a simplified 3D VAWT model is used where the supporting shaft and arms of the VAWT are excluded to reduce the needed computational capacity. Among the various endplate designs used in this study, the semi-circular inward endplate (ED3) with a diameter equivalent to 1.2 blade chord length showed the best improvement in the Cp which is by 7.45%, and 5.79% for at the TSRs of 2.19 and 2.58, respectively. The pressure difference on both sides of the blade was also examined. The results revealed that the endplate can prevent the flow from bypassing the blade tip, hence, preventing the occurrence of tip vortices while improving the aerodynamic efficiency near the blade tip, ultimately, improving the overall Cp of a VAWT.

Experimentation of helical flow path in an augmented wind turbine

The design of wind turbine blade for low wind speed area is historically experimented by many researchers and gives us different results with nearly same parameters. Also, augmentation of wind turbine improves power generation capacity in magnified manner. This paper deals with the fabrication and investigation of augmented wind turbine specially prepared with helical flow path. The experimentationdonewithcombinationofdifferentparameterssuchaswindspeed, pitch distance and angle of attack of wind flow.

Design and Optimization of a Small-Scale Horizontal Axis Wind Turbine Blade for Energy Harvesting at Low Wind Profile Areas

Wind turbine blades perform the most important function in the wind energy conversion process. It plays the most vital role of absorbing the kinetic energy of the wind, and converting it to mechanical energy before it is transformed into electrical energy by generators. In this work, National Advisory Committee for Aeronautics (NACA) 4412 and SG6043 airfoils were selected to design a small horizontal axis variable speed wind turbine blade for harvesting efficient energy from low wind speed areas. Due to the low wind profile of the targeted area, a blade of one-meter radius was considered in this study. To attain the set objectives of fast starting time and generate more torque and power at low wind speeds, optimization was carryout by varying Reynolds numbers (Re) on tip speed ratios (TSR) values of 4, 5, and 6. The blade element momentum (BEM) method was developed in MATLAB programming code to iteratively find the best twist and chord distributions along the one-meter blade length for each Re and tip speed ratio (TSR) value. To further enhance the blade performance, the twist and chord distributions were transferred to Q-blade software, where simulations of the power coefficients (Cp) were performed and further optimized by varying the angles of attack. The highest power coefficients values of 0.42, 0.43, and 0.44 were recorded with NACA 4412 rotor blades, and 0.43, 0.44, and 0.45 with SG6043 rotor blades. At the Re of 3.0 × 105, the blades were able to harvest maximum power of 144.73 watts (W), 159.69 W, and 201.04 W with the NACA 4412 and 213.15 W, 226.44 W, 245.09 W with the SG6043 at the TSR of 4, 5, and 6 respectively. The lowest cut-in speed of 1.80 m/s and 1.70 m/s were achieved with NACA 4412 and SG6043 airfoils at TSR 4. At a low wind speed of 4 m/s, the blades were able to harness an efficient power of 79.3. W and 80.10 W with both rotor blades at the TSR 4 and 6 accordingly.

Design, modeling and optimization of a renewable-based system for power generation and hydrogen production

Due to the environmental concerns caused by fossil fuels, renewable energy systems came into consideration. In this study, a renewable hybrid system based on ocean thermal, solar and wind energy sources were designed for power generation and hydrogen production. To analyze the system, a techno-economic model was exerted in order to calculate the exergy efficiency as well as the cost rate and the hydrogen production. The main parameters that affect the system performance were identified, and the impact of each parameter on the main outputs of the system was analyzed as well. The thermo-economic analysis showed that the most effective parameters on the exergy efficiency and total cost rate are the wind speed and solar collector area, respectively. To reach the optimum performance of the system, multi-objective optimization, by using genetic algorithm, was applied. The optimization was divided into two separate case studies; in case A, the cost rate and the exergy efficiency were considered as two objective functions; and in case B, the cost rate and the hydrogen production were assigned as two other objective functions. The optimization results of the case A showed that for the total cost rate of 30.5 $/h, the exergy efficiency could achieve 35.57%. While, the optimization of the case B showed that for the total cost rate of 28.06 $/h, the hydrogen production rate could reach 5.104 kg/h. Furthermore, after optimizing, an improvement in exergy efficiency was obtained, approximately 19%. • Proposing a novel integrated system based on ocean thermal, solar and wind energy. • Applying a comprehensive thermo-economic model to analyze the system. • Hydrogen production, exergy efficiency, and cost rate are the objective functions. • Enhancing the exergy efficiency 19% and the hydrogen production 1.14 kg/h.

Influence of Tropical Cyclones on Outdoor Wind Environment in High-Rise Residential Areas in Zhejiang Province, China

Tropical cyclones can cause tremendous harm to coastal areas. This research aims to explore the influence of tropical cyclones on outdoor wind environments in high-rise residential areas in the southeast coastal provinces of China, using Zhejiang Province as an example. It investigated four cities located in Zhejiang Province, including Taizhou and Wenzhou representing coastal cities, and Huzhou and Jiaxing representing inland cities, and collected data from 209 high-rise residential areas. Of these 209 samples, 131 high-rise residential areas with three typical spatial layouts (i.e., rows-style, free-style, and courtyard-style) have been selected for further studies. Numerical simulation was conducted to analyze the outdoor wind environment of these three types of high-rise residential areas, where the height of buildings was set as 26 floors (75 m), during tropical cyclones. Based on a comparison of the wind velocity at the horizontal planes of 1.5 m high and 10 m high, it was found that the spatial layouts of high-rise residential areas could mitigate the negative impact of tropical cyclones on the outdoor wind environment. Specifically, in the coastal cities, the courtyard-style layout led to a relatively small proportion of high wind speed areas (e.g., wind velocity above 14.4 m/s) in the high-rise residential areas; and in the inland cities, the free-style layout led to a relatively small proportion of high wind speed area in the high-rise residential area. In turn, to better cope with the tropical cyclones, it was suggested that the courtyard-style layout should be recommended for high-rise residential areas located in the coastal cities and the free-style layout should be recommended for high-rise residential areas located in the inland cities in coastal provinces.

Increasing the Energy Density and Power Ratio of a Staggered VAWT Wind Farm by Using the Rotor's Diameter as a Reference

The development of wind energy systems has achieved a higher technology readiness level for Horizontal Axis Wind Turbine (HAWT). Unfortunately, the HAWT is only suitable for high wind speed areas. The Vertical Axis Wind Turbine (VAWT) is considered the ideal model to utilize wind energy in the low wind speed region. However, VAWT has a lower power coefficient. Therefore, developing a VAWT wind farm can improve the overall energy density for power generation in the low wind speed region. In this study, staggered configuration for three turbine clusters is evaluated through numerical simulation and experimental tests. The pitch distance is set by using the rotor's diameter as a reference for placing the 3rd rotor at the second row. The turbulence intensity in the area wake superposition is highly affected by the position of the 3rd rotor. The flow characteristic indicates that the 3D layout has a high concentration at the front area of the 3rd rotor. It leads to higher achievement of power ratio for the clusters. The overall power ratio for 3D layout can achieve more than 0.9, whereas, at a speed 3 m/s, the highest power ratio is obtained at 1.0. The finding in this study can be set as an essential reference for developing a VAWT wind farm with a specific arrangement and improving the overall power density of the turbine clusters.

Validation of the CFOSAT Scatterometer Data With Buoy Observations and Tests of Operational Application to Extreme Weather Forecasts in Taiwan Strait

AbstractThe China‐France Oceanography Satellite (CFOSAT) launched in 2018 is equipped with the Chinese Scatterometer (CSCAT), which is designed to measure high‐precision sea surface wind fields in the global ocean. This study aims to validate the CFOSAT wind fields with moored buoy‐measured ground truth data from August 2019 to July 2021. The test area was chosen as the Taiwan Strait, which is an important navigation channel and rich fishery grounds in the East Asia. It is also a high wind speed area particularly during typhoons and cold air passages. Thus, the accurate wind forecasts are crucially needed to avoid capsizing and casualties. The validation results give the correlation coefficient (R2) of 0.92, the root mean square error (RMSE) of 1.75 m s−1, and the absolute deviation (AD) of 1.10 m s−1 for the wind speeds, and R2 of 0.96, the RMSE of 21.87°, and the AD of 1.17° for the wind directions. During cold air passages, the AD of wind speeds decreases to 1.03 m s−1, which is smaller than 1.17 m s−1 during typhoons. Using the CFOSAT wind as historical observation data for refining of forecast work was tested by model output statistics revised method. The test results show that the refined forecast results of wind fields are improved up to 4%–17%. Thus, the validation and test results obtained in this study indicate that updating satellite winds would have great contributions to refined forecasting of local sea surface winds.

SAF-Net: A spatio-temporal deep learning method for typhoon intensity prediction

A typhoon is a destructive weather system that can cause severe casualties and economic losses. Typhoon intensity (TI) is a measurement to evaluate its ruinous degree. Hence, typhoon intensity prediction is an important research problem and many methods have been proposed. However, most of the existing approaches have very limited capability to combine the 2D Typhoon Structure Domain-expert Knowledge (2D-TSDK) and the 3D Typhoon Structure Data-driven Knowledge (3D-TSDK) for the TI prediction. To address this issue, this paper proposes a spatio-temporal deep learning method named Spatial Attention Fusing Network (SAF-Net). The designed model aims to fuse the 2D-TSDK and the 3D-TSDK by developing a specific Wide & Deep framework. In the data-driven component, a special Spatial Attention (SA) module is designed to automatically select high-response wind speed areas and embedded into a three-branch CNN to exploit the 3D-TSDK. Then, the Wide & Deep framework integrates the 2D-TSDK and the 3D-TSDK for the TI prediction. Comprehensive experiments have been conducted on a real-world dataset, and the result shows that the proposed method outperforms state-of-the-art typhoon intensity prediction methods. The code is available in GitHub: https://github.com/xuguangning1218/TI_Prediction