Weather Situations Research Articles

Impact Analysis of Super Typhoon 2114 ‘Chanthu’ on the Air Quality of Coastal Cities in Southeast China Based on Multi-Source Measurements

The northward typhoon configuration along the southeast coast of China (TCN-SEC) is one of the key circulation patterns influencing the coastal cities in southeast China (CCSE). Here, we analyzed the air quality in CCSE during the high-incidence typhoon period from 2019 to 2021. Multi-source measurements were carried out to explore the impact of super typhoon 2114 ‘Chanthu’ on the air quality in CCSE. The results showed that the TCN-SEC and its surrounding weather situation had a favorable impact on the increase in pollutant concentration in CCSE, especially on the increase in O3 concentration. From 13 September to 17 September 2021, affected by the cyclonic shear in the south of super typhoon 2114 ‘Chanthu,’ the strong wind near the ground, stable relative humidity, strong precipitation, and the significantly reduced wind speed had a substantial effect on PM10, PM2.5, SO2, and NO2 concentrations. Calm and light air near the ground, weak precipitation, high daily maximum temperatures, and minimum relative humidity may provide favorable meteorological conditions for the accumulation of O3 precursors and photochemical reactions during the day, resulting in the daily peak values of O3 exceeding 160 μg/m3. The evolution of wind, relative humidity, and boundary layer height could play an important role in the variations in PM10 and PM2.5 concentrations by influencing pollutant accumulation or diffusion. It was suggested that the atmospheric structure of horizontal stability and vertical mixing below 1500 m could play a significant role in the accumulation and vertical distribution of ozone. The results highlight the important role of typhoons in the regional environment and provide a scientific basis for further application of multi-source observation data, as well as air pollution control.

Accuracy of radar-estimated precipitation in a mountain catchment in Slovakia

Abstract Accurate estimation of precipitation in mountain catchments is challenging due to its high spatial variability and lack of measured ground data. Weather radar can help to provide precipitation estimates in such conditions. This study investigates the differences between measured and radar-estimated daily precipitation in the mountain catchment of the Jalovecký Creek (area 22 km2, 6 rain gauges at altitudes 815–1900 m a.s.l.) in years 2017–2020. Despite good correlations between measured and radar-based precipitation at individual sites (correlation coefficients 0.68–0.90), the radar-estimated precipitation was mostly substantially smaller than measured precipitation. The underestimation was smaller at lower altitude (on average by –4% to –17% at 815 m a.s.l.) than at higher altitudes (–35% to –59% at 1400–1900 m a.s.l.). Unlike measured data, the radar-estimated precipitation did not show the differences in precipitation amounts at lower and higher altitudes (altitudinal differences). The differences between the measured and radar-estimated precipitation were not related to synoptic weather situations. The obtained results can be useful in preparation of more accurate precipitation estimates for the small mountain catchments.

Design of Road surveillance system for low visibility, bad weather and emergency situations

Traffic accidents are major causes of death and disability worldwide. Lack of visibility due to Fog and darkness are main causes of accidents. These accidents may turn into fatalities and long traffic jams if proper actions are not taken. In this paper we initially review the accident data and scenarios and then propose a road surveillance system in low visibility and emergency. This proposed system that mainly performs three tasks as turning ON road-alert lights, if a vehicle unexpectedly stops on road-side due to any technical fault or accidents. This will help upcoming vehicles to have prior information of unexpectedly stopped vehicle that will reduce accident due to vehicles collision, other side turning ON road-street lights automatically with darkness detection. This will reduce accident by improving visibility in night or fog and Send Emergency Notification to the control room if vehicle remain still for more than 20 seconds. This will help in taking proper and timely action in case of emergencies avoiding fatalities and traffic jams for hours. The system is based on Arduino and GSM module and tested for each of the three scenarios (fog, darkness, emergency). The result shows that the proposed system worked efficiently in each condition and can be used to improve traffic safety.

Look-Alike Detection using Deep Convolutional Neural Networks for Ocean Oil Spills

The marine and coastal ecosystems are placed in grave danger whenever there is a spill of oil. For the sake of protecting future generations, the authorities need a dependable mechanism that allows them to respond quickly and effectively to oil spills. As a result of their consistency also effectiveness in a wide series of weather and light situations, synthetic aperture radar (SAR) sensors are frequently utilized for this determination. The dark spots that are typically associated with oil spills remain easy for SAR sensors to spot, but it may be difficult to differentiate them from other objects or phenomena. A great number of methods have been outlined in order to automatically differentiate and categories these dark spots. The results are frequently unable to be compared with one another because of the heterogeneous nature of the data collected. It is often difficult to fine-tune settings or extract meaningful information because SAR images remain often classified through a solo label that is applicable to the entire picture. Because of this, the process of dealing with the images is made more difficult. The Random Forest Classifier and deep convolutional neural networks (also known as DCNNs) are examples of approaches that have been suggested as potential workarounds for these constraints. In addition, we make available to the public a search-and-rescue photo library that might serve as a standard for the growth of upcoming oil spill recognition tools. On the dataset that was provided, a number of well-known DCNN segmentation models as well as the Random Forest technique are evaluated and compared with one another. When compared to other approaches, it was discovered that Random Forest had the best test correctness and the quickest implication time. Additionally, the dataset is used to explore and explain the challenges of the provided issue, which in this case is identifying between genuine and fabricated oil spills. These investigations and demonstrations are carried out with the help of the dataset. According to their training and evaluation on the dataset that was provided to them, DCNN segmentation models, when paired with a Random Forest classifier, did quite well when attempting to detect oil spills. The novel method is anticipated to be of significant use in subsequent research pertaining to the identification of oil spills and the dispensation of SAR images.

Lived Experiences of the Philippine Coast Guards Personnel in Performing their Functions

The Philippine Coast Guard Personnel played a vital role on both the land and water, with its main responsibilities being the protectors of marine resources, the environment, and the lives and property at sea, as well as aiding in the enforcement of maritime laws under Philippine Jurisdiction. Their experiences in carrying out their duties include a variety of activities and problems. Each time they face harsh weather and tough situations in order to save lives and respond to distress calls, they risk their personal safety, witness difficult circumstances, and reap the benefits of supporting others in need throughout their lives. Their training, expertise, and commitment enable them to overcome challenges and carry out their responsibilities as Philippine coast guardians. The study explored lived experiences of the Philippine Coast Guard Personnel in performing their functions. The study included ten (10) personnel of Philippines Coast Guards participants of the study. This study was conducted during face to face interviewed. Results showed that the four (4) main themes based on the responses of the respondents such as: Family Centeredness, Sense of Service, Performing Challenging Duties and Responsibilities and Dealing with the Challenges. The personnel of Philippine Coast guard were continuously performing their duties and responsibilities for the safety of community travelers and living species in sea. Thus, safeguarding the sea and keeping citizens safe in voyage is evident that the personnel of Philippine coast guard were diligent in doing their duties and responsibilities.

Deep Learning–Based Parameter Transfer in Meteorological Data

Abstract Numerical simulations in Earth-system sciences consider a multitude of physical parameters in space and time, leading to severe input/output (I/O) bandwidth requirements and challenges in subsequent data analysis tasks. Deep learning–based identification of redundant parameters and prediction of those from other parameters, that is, variable-to-variable (V2V) transfer, has been proposed as an approach to lessening the bandwidth requirements and streamlining subsequent data analysis. In this paper, we examine the applicability of V2V to meteorological reanalysis data. We find that redundancies within pairs of parameter fields are limited, which hinders application of the original V2V algorithm. Therefore, we assess the predictive strength of reanalysis parameters by analyzing the learning behavior of V2V reconstruction networks in an ablation study. We demonstrate that efficient V2V transfer becomes possible when considering groups of parameter fields for transfer and propose an algorithm to implement this. We investigate further whether the neural networks trained in the V2V process can yield insightful representations of recurring patterns in the data. The interpretability of these representations is assessed via layerwise relevance propagation that highlights field areas and parameters of high importance for the reconstruction model. Applied to reanalysis data, this allows for uncovering mutual relationships between landscape orography and different regional weather situations. We see our approach as an effective means to reduce bandwidth requirements in numerical weather simulations, which can be used on top of conventional data compression schemes. The proposed identification of multiparameter features can spawn further research on the importance of regional weather situations for parameter prediction and also in other kinds of simulation data.

Estimation of near-surface ozone concentration and analysis of main weather situation in China based on machine learning model and Himawari-8 TOAR data

Ozone (O3) is an important greenhouse gas in the atmosphere. Stratospheric ozone protects human beings, but high near-surface ozone concentrations threaten environment and human health. Owing to the uneven distribution of ground-monitoring stations and the low time resolution of polar orbiting satellites, it is difficult to accurately evaluate the refinement and synergistic pollution of near-surface ozone in China. Besides, atmospheric circulation patterns also affect ozone concentrations greatly. In this study, a new generation of geostationary satellite is used to estimate the hourly near-surface ozone concentration with a spatial resolution of 0.05°. First, the Pearson correlation coefficient and maximum information coefficient were used to study the correlation between the top of atmospheric radiation (TOAR) of Himawari-8 satellite and O3 concentration; seven TOAR channels were selected. Second, based on an interpretable deep learning model, the hourly ozone concentration in China from September 2015 to August 2021 was obtained using the TOAR-O3 model. Finally, the self-organizing map method was used to determine six major summer weather circulation patterns in China. The results showed that (1) the near-surface O3 concentration can be accurately estimated; the R2 (RMSE: μg/m3) values of the daily, monthly, and annual tenfold cross validation results were 0.91 (12.74), 0.97 (5.64), and 0.98 (1.75), respectively. The feature importance of the model showed that the temperature, TOAR, and boundary layer height contributed 38 %, 22 %, and 13 %, respectively. (2) The O3 concentration showed obvious spatiotemporal difference and gradually increased from 10:00 to 15:00 (Beijing time) every day. In most areas of China, O3 concentration had increased significantly. (3) The O3 concentration in northern China was the highest under the circulation pattern of the Meiyu front over the Yangtze River Delta, while in southern China, it was the highest under the circulation pattern of the northeast cold vortex controlling most of China.

Comparison of Urban Heat Island Diurnal Cycles under Various Atmospheric Conditions Using WRF-UCM

The rapid growth of urbanization significantly influences local atmospheric conditions and life quality of residents living in urban areas by creating a localized phenomenon known as an urban heat island. Urban heat island characteristics are strongly formed by prevailing atmospheric conditions influencing their magnitude and intensity. In this study, we used the Weather Research and Forecasting model to investigate the effects of different airmass and windspeed characteristics on the diurnal cycle of the urban heat island phenomenon during four real weather situations presenting typical scenarios with clear sky conditions. The scenarios consisted of warm dynamic and non-dynamic situations and cold dynamic and non-dynamic situations identified with respect to temperature profiles, humidity and wind speed. The comparison of urban heat island intensity between all four scenarios showed a significant effect of wind speed on urban heat island characteristics and intensity as well as the role of humidity and airmass temperature in diurnal changes. The results showed that urban heat island is best defined by conditions with strong radiative heating and weak wind speed regardless of temperature. Air humidity appears to have significant influence on UHII with cold non-dynamic situations in dry air showing a very high UHII amplitude during the daylight period and high intensity during the night. The comparison of warm and cold dynamics situations showed the influence of vertical heat exchange with strong mixing of air between near-surface levels and lower levels of the troposphere with a mild effect on UHII during the warm situation and strong influence on decreasing UHII during the cold scenario.

Artificial intelligence and cloud-based Collaborative Platforms for Managing Disaster, extreme weather and emergency operations

Natural disasters are often unpredictable and therefore there is a need for quick and effective response to save lives and infrastructure. Hence, this study is aimed at achieving timely, anticipated and effective response throughout the cycle of a disaster, extreme weather and emergency operations management with the help of advanced technologies. This study proposes a novel, evidence-based framework (4-AIDE) that highlights the role of artificial intelligence (AI) and cloud-based collaborative platforms in disaster, extreme weather and emergency situations. A qualitative approach underpinned by organizational information processing theory (OIPT) is employed to design, develop and conduct semi-structured interviews with 33 respondents having experience in AI and cloud computing industries during emergency and extreme weather situations. For analysing the collected data, axial, open and selective coding is used that further develop themes, propositions and an evidence-based framework. The study findings indicate that AI and cloud-based collaborative platforms offer a structured and logical approach to enable two-way, algorithm-based communication to collect, analyse and design effective management strategies for disaster and extreme weather situations. Managers of public systems or businesses can collect and analyse data to predict possible outcomes and take necessary actions in an extreme weather situation. Communities and societies can be more resilient by transmitting and receiving data to AI and cloud-based collaborative platforms. These actions can also help policymakers identify critical pockets and guide administration for their necessary preparation for unexpected, extreme weather, and emergency events. • AI and Cloud technologies for disaster and emergency operations management. • OIPT view to address the emergency situations through AI and cloud technologies. • Real-time data integration through AI and cloud technologies in emergency situations. • 4-AIDE framework of readiness, response, recover and re-evaluate for better disaster management.

Association of weather conditions and the day with extreme number of deliveries with spontaneous onset in a tertiary referral perinatal center

Background and purpose: The effect of weather on people’s well-being and health has been previously noticed and has been a subject of interest for medical professionals and laypeople throughout human history. There are many studies connecting gynecology and obstetrics with weather, some of them investigating the weather and physiological processes such as onset of labor.Materials and methods: In this paper we tried to find relationship between weather conditions and the day with extreme number of deliveries with spontaneous onset (contractions and/or rupture of membranes) in a tertiary referral perinatal center. It is still debatable whether we could connect the weather conditions with actual childbirth.Results: A case analysis shows that there could be a connection between the development of the weather situation and the extreme number of deliveries with spontaneous onset.Conclusion: Unfavorable biometeorological conditions were the result of weather conditions that affect people. In our case there was strong cold advection during the analyzed period, especially on the day with an extreme number of deliveries with spontaneous onset, and significant drop of barometric pressure.

A Study of Air Pollution Episode Over Delhi National Capital Region, India During October 29, 2016- November 03, 2016

Northern India has an increase in surface air pressure during winter; this causes the atmosphere to clearout, resulting in the production of significant surface inversions. High-pressure zones are also calm andwind direction changes only when a western disturbance enters the area. The wind is substantially strongerwhen there is a low-pressure system. As a result, the pollutants in northern India are regulated by a sequenceof high- and low-pressure weather situations. October 29, 2016, was when the wind direction unexpectedlyshifted from NW to NE and then to N. Wind speeds dropped from 1-6 km/h to near-calm during this time,which indicated the passing of a high-pressure zone, based on the readings of the pressure gauges. Pollutionfrom northern states was brought into the national capital region by a combination of low wind speed andthe wind’s direction. NCR was clogged with high levels of particulate matter during Diwali celebrations,which were exacerbated by heavy traffic, factory production, and firecrackers. The grave situation persisteduntil the wind shifted directions and increased its velocity. On the monostatic acoustic radar installed atREMTech, Shamli, detected the persistence of a substantially raised layer over the region, implying thatvertical mixing of pollutants was severely harmed or limited. As a result of this state, pollutants accumulated,resulting in a severe bout of air pollution. The findings of this study details an incident of such air pollution happened over Delhi NCR.

Features and Evolution of Autumn Weather Regimes in the Southeast China

Autumn is the transitional season when the atmospheric circulation pattern changes from summer to winter. The temperature and precipitation in Southeastern China in autumn are significantly influenced by the change in circulation patterns, and both show significant uniqueness. The clustering method can be used to observe the changes of circulation patterns in detail and to observe and analyze the transition from warm to cold seasons from a detailed view of the daily circulation pattern perspective. This method may have important research implications on how to study the generation and dissipation of extreme weather events. The Self-Organizing Maps (SOM) method is used to a 500 hPa geopotential height and 850 hPa wind and sea level pressure for 1981–2020 to identify the characteristic weather patterns (WTs) in autumn (September–November) over Southeastern China. Characteristics of the captured WTs are also analyzed in terms of the distribution characteristics of weather patterns, occurrence frequency, typical progression, precipitation and extreme precipitation (EP), temperature and extreme high temperature (EHT), and the relationship with atmospheric teleconnection. Nine WTs were identified in autumn, which represents a series of weather situations consisting of troughs and ridges. On this basis, these WTs were used to carry out the differentiation of seasonal differences between early and late autumn. The maximum mean and extreme precipitation occur in several early season patterns (WT1, WT2, WT4, and WT7). It is highly likely that extremely high temperatures occur in the WT1 and WT2 patterns. The most common progression between WTs is WT7−WT1−WT2−WT4 in the early season. This seasonality allows us to distinguish between early and late seasons based on daily weather types. A preliminary trend analysis suggests that patterns in the early season occur more frequently and last longer in the early season, and patterns in the late season occur less frequently and later. That is, the longer cool season pattern is shifting to the shorter warm season pattern. In addition, the persistence of both cool and warm patterns increased during 2001–2020 relative to 1981–2000, and the risk of both flooding and drought occurrence is on the rise.

The summer 2021 Switzerland hailstorms: weather situation, major impacts and unique ­observational data

On 28 June and 8 July 2021, all the ingredients for severe convection were present and resulted in severe hailstorms over Switzerland, causing historic damage, as recorded by several insurance companies. The return periods of those events locally exceeded 70–100 years according to the new Swiss hail climatology. A coherent and comprehensive picture of those hailstorms is given by a unique combination of hail observing systems: automatic hail sensors, crowdsourced reports and radar-based algorithms. (Figure: Sturmarchiv Schweiz/www.sturmarchiv.ch © Martin Hotz, Steinhausen.)

Development of MVMD-EO-LSTM Model for a Short-Term Photovoltaic Power Prediction

The accuracy and stability of short-term photovoltaic (PV) power prediction is crucial for power planning and dispatching in a grid system. For this reason, the multi-resolution variational modal decomposition (MVMD) method is proposed to achieve multi-scale input features mining for short-term PV power prediction. Here, the MVMD combined with Spearman extracts correlation features of the weather data. An equilibrium optimizer (EO) is integrated with MVMD to achieve optimal values of the long short-term memory (LSTM) parameters. Firstly, the correlation of input features is determined and selected by Spearman. The MVMD model is used to mine the high correlation features of solar radiation and conduct cross-correlation analysis to extract input feature components. Secondly, the similar weather days of the sample set are classified to ensure a good adaptability in different weather situations. Finally, the high correlation features are introduced into the photovoltaic power prediction model of EO optimized LSTM. Performance analysis using actual output power data from a PV plant shows that the proposed MVMD feature extraction method can effectively mine correlation features to achieve an optimized dataset under different seasons. Compared with the gray wolf and particle swarm optimization algorithms, the proposed model has a better optimization performance in a low discrimination of input feature decomposition components and low correlation with output power.

Growth and productivity of gobhi sarson (Brassica napus) cultivars under diverse sowing environments

The sowing date of oilseed [winter (rabi) season] crop changes from year to year based on the harvesting of the preceding sown (rainy (kharif) season) crop in the northern regions of India, which exposes the crop to varying weather situations. So, to investigate the impact of sowing environments on growth, yield attributes and yield of gobhi sarson (Brassica napus L.) cultivars, an experiment was conducted with the 3 sowing environments and 3 cultivars under sub-tropical conditions of Jammu during the winter (rabi) season of 2018–19 and 2019–20. The pooled data of two years revealed that the early sown gobhi sarson crop recorded significantly highest growth parameters (plant height, leaf area index and dry matter accumulation) and yield attributes (number of primary and secondary branches/plant, siliquae/ plant and seeds/siliqua) as compared to the delayed sown crop. With the result (source sink relationship), significantly higher seed yield was recorded under the first sowing environment (1758 kg/ha) sown crop, while significant reduction to the extent of 27.66% and 66.95% was recorded with delayed sowing, D2 (1377 kg/ha) and D3 (1053 kg/ha), respectively. The cultivar GSL-1 recorded significantly higher plant height, LAI, dry matter, primary and secondary branches/plant, siliquae/plant, seeds/silique, followed by ONK-1 and DGS-1, resulting in significantly higher yield of GSL-1 (1493 kg/ha) over DGS-1 (1262 kg/ha) but at par with ONK-1 (1432 kg/ha).

Explosive COVID-19 outbreak in a German nursing home and the possible role of the air ventilation system

BackgroundMost COVID-19 outbreaks in nursing homes are explained by transmission of SARS-CoV-2 from nurses or visitors.Methods and resultsWe describe an outbreak with 64 of the 67 residents identified as COVID-19 cases within two weeks (34 in nursing block 1, 30 in nursing block 2), at least 32 of them had relevant symptoms of COVID-19. Thirteen of the residents’ deaths were associated with COVID-19. In addition, 27 of approximately 60 staff members were identified as COVID-19 cases, 23 of them had relevant symptoms. In none of the samples from residents or staff was a mutation of SARS-CoV-2 detected. Quarantine of the residents was already in force at the beginning of the outbreak. A common source among the staff was considered to be unlikely because the two nursing home blocks had no staff rotation and the staff had to wear FFP2 masks during contact with residents. Three months after the outbreak the RNA of SARS-CoV-2 was detected on 14 of 39 sampled indoor surfaces of the air ventilation system with Ct values between 34.9 and 41.9, but only at the air supply in the corridor (11 of 24 samples) and the air overflow in the door between the corridor and the residents’ rooms (three of 11 samples) but not at the air exhaust in the residents’ bathrooms.ConclusionsThe air ventilation system and an inversion weather situation three days before the first confirmed case may have enhanced viral spread inside the nursing home assuming that a common source with a high viral load had existed at the time of outbreak.

Weather Forecasting Method from Sensor Transmitted Data for Smart Cities Using IoT

India experiences severe weather events around the year. Severe thunderstorms occur during the premonsoon season (March-May), cyclonic storms occur over the Bay of Bengal/Arabian Sea during the premonsoon and postmonsoon seasons (October-December), and heavy rainfall occurs during the monsoon season (June-September). It causes a lot of damage to property and the lives of humans. Smart urban areas aim to improve residents’ personal satisfaction by utilizing data about urban scale procedures separated from heterogeneous information sources gathered on city-wide arrangements using sensors. The Internet of Things (IoT) is an empowering concept for forecasting weather situations on an urban scale. A multisource detecting framework influences IoT innovation to accomplish city-scale detection of climatic changes and forecasts them to citizens in a smart city. The existing models gather the data in an unstructured process that need to be structured. This is a complex task that needs to be reduced. The proposed model gathers data from various regions in a city and uses it to identify the weather regions. A novel approach is introduced to detect this climate information gathered by utilizing various sensors arranged in a city. The information gathered from the sensors is thoroughly examined to see if there is any inconsistency in weather reports in any of its key hubs, and cautions are activated to the city for prompt actions. In this proposed work, an efficient method for weather casting using an IoT mechanism is introduced, and the results state that the proposed method is effective in terms of accuracy and speed when contrasted with the traditional methodologies.

A Novel Fuzzy Based Solar Irradiation Prediction Approach for Selection of Solar Panels in Hisar (Haryana)

This paper proposes an ANN-based framework for forecasting the daily solar irradiance with the meteorological data collected over 10 years for Hisar city, located in Haryana. Meteorological information like temperature, relative humidity, and average wind speed is used as input variables for estimating sun irradiation in order to obtain high accuracy under various weather situations. Meteorological information is used as input variables for estimating the solar irradiation in order to obtain acceptable accuracy in a variety of weather circumstances. In addition, it was suggested in this paper that the Artificial Neural Network (ANN) model be trained using the Harris Hawks Optimization (HHO) Algorithm. The following task is to choose the best solar panel (SP) out of the thousands that are currently on the market. Due to a crucial component like weather, choosing the finest SP is important. Hence, this paper introduces a score-based model for selecting the best panel by considering the different criteria of the eleven SPs that are used as test panels. The annual power produced by panel P11 is 79.67% better than panel P1. Finally, based on the average power generated, the proposed panel P11 is regarded as the best panel when compared to other panels.

A Thunderstorm Gale Forecast Method Based on the Objective Classification and Continuous Probability

Potential prediction is an important research content of thunderstorm gale weather forecast, and it is still a challenge because the environmental field of thunderstorm gale presents different characteristics under different weather conditions. Using the 12-year thunderstorm gale data of Hubei province in central China and the reanalysis data of National Center for Environmental Prediction (NCEP), this study analyzed the percentile distribution of environmental physical quantities of thunderstorm gale, and the continuous probability method was adopted to establish the probability forecast models of thunderstorm gale in four different types of weather situation, which are in the rear of trough type, in front of trough type, in the periphery of the western Pacific subtropical high type and easterly airflow type. Finally, probability prediction was realized by objective classification criterion in operation. The results show that the method based on objective classification and continuous probability can significantly improve the probability of thunderstorm gale detection, and also reduce the missing alarm rate of thunderstorm gale. Moreover, the quantitative test of 16 weather processes under four types of weather situations also shows that the continuous probability method has a higher probability of detection than the bisection method, and significantly reduces the missing alarm of extreme wind by the bisection method.

Memory-based Transformer with shorter window and longer horizon for multivariate time series forecasting

Multivariate time series forecasting is an important problem that spans many fields. One challenge of this problem is the complex and non-linear interdependence between time steps and different variables. Recent studies have shown that Transformer has potential in capturing long-term dependencies. However, in the field of time series forecasting, Transformer still has some problems to solve, such as prediction fragmentation and insensitivity to data scale. In addition, traditional forecasting models often require a large amount of input data to support the training of the model when predicting long-term data. However, it is hard to provide sufficient time series input data due to equipment damage or weather situation. To solve these limitations, a memory-based Transformer with shorter window and longer horizon is proposed, called SWLHT. It uses the memory mechanism to make the model no longer only rely on a single input, but can combine the previous forecast results to assist in capturing long-term dependencies, thereby avoiding the requirement of excessively long input sequence. Furthermore, the memory mechanism can alleviate the prediction fragmentation to some extent. The experimental results and comparison of baselines on several real-world multivariate time series datasets have verified the effectiveness of the proposed model.