Local Forecasts Research Articles

Regional, local and current forecast of the impact hazard of coal seam sites based on seismic monitoring

Relevance. Insufficient validity of the quantitative criteria of the regional, local and current forecast of the impact hazard of the coal seams being developed, the absence of algorithms linking the use of geophysical and direct geomechanical methods in forecasting. Aim. To develop a comprehensive method of geodynamic forecasting, including determining the extent of an impact-prone area identified by the results of a regional forecast based on the registration data of the GITS seismic monitoring system, for conducting local and current forecasts of impact hazard by regulatory methods. Object. An array of rocks of the excavation column 4-1-5-7 of the Osinnikovskaya mine during mining operations in the zone of a group of tectonic disturbances and the intersection of advanced workings. Methods. Regional geomechanical forecast by the geophysical method of seismic activity registration (GITS system), local (current) forecast by natural electromagnetic radiation methods (Angel-M equipment) and the output of drilling fines. Results. The paper introduces the results of complex studies in the excavation column 4 1-5-7 of the Osinnikovskaya mine to establish a correlation between seismic activity and the parameters of the stress-strain state of the array. The authors have introduced the concept of integral indicators of the considered parameters IF and Ikσ. They take into account the area of the zone for the i-th range according to the complex parameter of seismic activity F and the coefficient of vertical stresses in the roof of the formation Kσ. It is established that the informative value of the integral indicators is almost twice as high as the informative value of the initial parameters. Experimental nomograms и are presented to determine the category "DANGEROUS/NON-DANGEROUS" for the regional forecast of impact hazard according to the GITS seismic monitoring system. The authors developed a general algorithm for predicting the impact hazard during clean-up operations. This algorithm uses an integrated approach that includes a regional forecast for recording seismic activity within the mine field and local forecasting methods, both geophysical and direct, to clarify the boundaries of seismic energy release zones and confirm the category of impact hazard. Thу paper introduces the example of determining the width of the zone of increased seismic energy release in the mine closest to the seismically active zone. The dynamics of the occurrence, development and decrease of the seismic energy release zone in the considered areas in the mine workings is shown. The authors give the examples of the implementation of a local forecast of impact hazard in previously established potentially dangerous areas by methods of recording natural electromagnetic radiation of rocks and by the output of drilling fines.

Deep learning for automated trash screen blockage detection using cameras: Actionable information for flood risk management

ABSTRACT Trash screens are used to prevent debris from entering critical parts of rivers. However, debris can accumulate on the screen and generate floods. This makes their monitoring critical both for maintenance and flood modeling purposes (e.g., local forecasts may change because the trash screen is blocked). We developed three novel deep learning methods for trash screen maintenance management consisting of automatically detecting trash screen blockage using cameras: a method based on image classification, a method based on image similarity matching, and a method based on anomaly detection. To facilitate their use by end users, these methods are designed so that they can be directly applied to any new trash screen camera installed by the end users. We have built a new dataset of labeled trash screen images to train and evaluate the efficiency of our methods, in terms of both accuracy and implications for end users. This dataset consists of 80,452 trash screen images from 54 cameras installed by the Environment Agency (UK). This work demonstrates that trash screen blockage detection can be automated using trash screen cameras and deep learning, which could have an impact on both trash screen management and flood modeling.

The performance of Climate Information Service in delivering scientific, local, and hybrid weather forecasts: A study case in Bangladesh

Access to reliable and skillful Climate Information Service (CIS) is crucial for smallholder farmers in Bangladesh to mitigate the impacts of rainfall variability and extremes. This study aims to systematically evaluate the performance of CIS in providing Scientific Forecast (SF) and Local Forecast (LF) to smallholders in Bangladesh. The results were then compared with farmers’ perceptions of the forecast accuracy. Additionally, the skill of a simple hybrid forecast (HF), which is an integrated system of SF and LF, was assessed using the ERA5 and ground observation datasets as benchmarks. The SF and LF data were obtained from the meteoblue hindcast and from the interview, respectively. The results indicate that, overall, LF exhibits slightly higher skill compared to SF when evaluated against the ERA5 dataset. The forecast performance, however, declines by almost half when the ground-based observations are used, associated with high false alarms. Farmers, on the other hand, perceived SF to possess superior performance compared to LF. This study demonstrates that combining the SF and LF into a simple HF yields higher forecast skill than either individual forecast, highlighting the importance of HF to deliver a reliable and trustworthy weather forecast.

Population Fusion Transformer for Subnational Population Forecasting

Forecasting the subnational population accurately is needed for sustainable development, including planning for the future, allocating resources, or providing health services. Two approaches are used for forecasting subnational populations: local forecasting where a model is trained for each area, and global forecasting, where one model is trained with all areas. Local forecasting (e.g., statistical models) is limited to capturing the population growth patterns in a single area. Machine learning models, such as the light gradient boosting model (LGBM), are considered a more suitable approach for global forecasting, but it is limited to one-step predictions, leading to error accumulation. Also, combining several models into one ensemble model are used which helped in reduce forecasting errors. However, the nature of population growth is nonlinear, and there is a need to reduce error accumulation. This study overcomes these issues and proposes a population fusion transformer (PFT) as a global forecasting model for population forecasting, which outputs multi-step predictions. The PFT is based on a temporal fusion transformer (TFT) proposing a novel deep gated residual network (DGRN) block to capture data nonlinearity. This study also incorporates the proposed PFT model into various ensemble models to reduce forecasting errors using different prediction and learning approaches. The proposed models are applied to four subnational population datasets from several countries. The PFT model outperforms the LGBM and TFT with lower forecasting errors in three and two datasets. More importantly, combining the PFT with other models in ensemble models reduced errors further.

Business unit controllers' credibility and the hardening of local forecasts

AbstractFocusing on multidivisional companies, this paper analyzes the hardening of local forecasts at the intersection of business units (BUs) and the corporate finance function. It investigates how BU controllers, accountable to both local management and the corporate finance function, seek to establish themselves as competent and trustworthy forecasters vis‐à‐vis their functional superiors. Drawing on Goffman's dramaturgical sociology, we demonstrate how these encounters constitute episodes in a multiperiod hardening game feeding into the management of forecast quality and anticipatory control. We illustrate how expressive performances of their competence and trustworthiness are vital for BU controllers to manage vertical information flows between the local and the corporate level and for aligning corresponding interests. Convincing performances can reinforce BU controllers' status as stewards of the forecasting process and help to maintain a “truce” between the local and the corporate level, assuring corporate controllers that the unit's future is under control.

Organizational and legal contexts of cooperative security: from global trends to local forecasts

Organizational and legal contexts of cooperative security: from global trends to local forecasts

Forecast of oil and gas reservoirs of the Sea of Okhotsk based on interpretation processing of seismic material

Background. This article analyses the results of geological and geophysical studies conducted in the Okhotsk Sea area with the purpose of developing methods for processing seismic information in combination with other data to solve the problems of forecasting hydrocarbon deposits.Aim. Solving problems of forecasting hydrocarbon deposits through the development of a structural and technological scheme for prospecting and evaluation of a geological and geophysical model to identify images of oil and gas deposits.Materials and methods. The research methodology included selection of the location of points for simulating “vertical wells” in order to analyze the geodynamics of the medium; multiplex parametric transformations of wave fields taking into account diagnostic significant attributes and building a multiparametric cluster; selection of oil and gas deposits; formation of images of oil and gas deposits; local forecast of improved reservoir properties and saturation; refinement of the geometry of target surfaces with forecast data in actual points (vertical wells); forecast of the area of target surfaces; spatial mapping of the largest hydrocarbon deposits.Results. Forecasting of oil and gas prospects in the Okhotsk Sea in areas where there is no deep drilling can be carried out using the following technologies: detailing the structural and tectonic structure of hydrocarbon traps; multi-feature prediction of hydrocarbon deposits; decomposition of the dispersion of the wave field; low-frequency resonance of seismic emission geodynamic noise; selection of oil and gas-containing objects.

DL[formula omitted]F: A Deep Learning model for the Local Forecasting of renewable sources

This work presents the first version of DL2F, a powerful Deep Learning model specifically designed to forecast four essential weather variables that influence solar power potential: Global Horizontal Irradiance (GHI), temperature, atmospheric pressure, and relative humidity. The model uses a time series for each variable as input and the forecast provided by the well-known Fifth-Generation Mesoscale Model (MM5) system. Then, it generates new predictions for each variable and the subsequent 3 days. The suggested approach improves the MM5 system’s accuracy, effectively mitigating its tendency to misestimate real data in some specific seasons. In essence, DL2F is a regressor able to refine the forecasts generated by the MM5 system and improve its overall precision. The model’s architecture is based on a set of recurrent neural networks of the Gated Recurrent Unit (GRU) type.

Localization patterns of gold mineralization in the Phuoc Son area, Central Vietnam

Background. In this paper, the authors aim to reveal the localization patterns of gold mineralization in the Phuok Son region of Central Vietnam, which formed a Proterozoic-Cambrian metamorphic complex in the zone of the Late Paleozoic Benzhang–Queshon granite block.Aim. To determine the localization patterns of gold mineralization in the Phuoc Son area, Central Vietnam.Materials and methods. The research was based on geological materials, as well as mine working and appraisal well data. The material composition of the materials was studied using 15 transparent sections and 18 polished sections. In total, 127 samples were subjected to atomic absorption spectroscopy.Results. Within the Phuoс Son gold ore field in Central Vietnam, the site of the Baidat and Baigo mines located in the zone of reverse-shear termination was investigated. Here, the gold ore bodies are represented by quartz-sulfide veins subparallel to schistosity. The ore-bearing zones are layered zones of tectonic faults, which formed at the end of the reverse-shear period and complicated the western limbs and locks of the anticlines. These zones were formed under regional sublatitudinal horizontal stresses. The horizons of carbonaceous quartz-sericite schists are of great importance in ore control. There are steeply dipping post-ore faults that displace ore bodies. Two gold mineralization stages can be distinguished. The early stage was characterized by a wide development of sphalerite and a smaller amount of gold. The later stage was associated with the main precipitation of gold and formation of galena. According to the geochronological data on the argon isotopy in biotite, the gold mineralization at Baidat and Baigo was probably formed in the Triassic (250–200 Ma ago), at the stage of extinction of the regional metamorphism of the Indosinian orogenyConclusion. The results obtained should be used when conducting local forecasting of gold deposits in Central Vietnam.

Литологическая характеристика предгребенского несогласия в разрезе ручья Сизимцелебейшор поднятия Чернова

A lithological description of natural outcrops of Upper Silurian rocks in the section of the Sizimcelebeyshor creek of the Chernov uplift has been presented. Based on the results of the studies, the sedimentation gap at the end of Gerd'yu superhorizon is characterized. We consider breccias and interlayers of red-green plastic eluvial clays associated with the unconformity as occurrences of the Silurian paleocarst. This factor should be taken into account in the local forecast of zones with improved reservoir properties for carbonate rocks. The conducted studies allow significantly replenishing the sedimentological and chronostratigraphic characteristics of this area of work, which in the future can contribute to solving problems of regional correlation, paleofacial reconstructions and optimization of geological exploration.

Meteorological Conditions of Extreme Heavy Rains over Coastal City Mumbai

Abstract The city of Mumbai, India, frequently receives extreme rainfall (>204.5 mm day−1) during the summer monsoonal period (June–September), causing flash floods and other hazards. An assessment of the meteorological conditions that lead to these rain events is carried out for 15 previous cases from 1980 to 2020. The moisture source for such rain events over Mumbai is generally an offshore trough, a midtropospheric cyclone, or a Bay of Bengal depression. The analysis shows that almost all of the extreme rain events are associated with at least two of these conditions co-occurring. The presence of a narrow zone of high sea surface temperature approximately along the latitude of Mumbai over the Arabian Sea can favor mesoscale convergence and is observed at least 3 days before the event. Anomalous wind remotely supplying copious moisture from the Bay of Bengal adds to the intensity of the rain event. The presence of midtropospheric circulation and offshore trough, along with the orographic lifting of the moisture, give a unique meteorological setup to bring about highly localized catastrophic extreme rainfall events over Mumbai. The approach adopted in this study can be utilized for other such locales to develop location-specific guidance that can aid the local forecasting and emergency response communities. Further, it also provides promise for using data-driven/machine learning–based pattern analysis for developing warning triggers. Significance Statement We have identified the meteorological conditions that lead to extreme heavy rains over Mumbai, India. They are that 1) at least two of these rain-bearing systems, offshore trough, midtropospheric circulation, and Bay of Bengal depression moving north-northwestward are concurrently present, 2) an anomalous high SST gradient is present along the same latitude as Mumbai, and 3) the Western Ghats orography favors the rainfall extreme to be highly localized over Mumbai.

Random Forest Approach for Improving Nonconvective High Wind Forecasting across Southeast Wyoming

Abstract High winds are one of the key forecast challenges across southeast Wyoming. The complex mountainous terrain across the region frequently results in strong gap winds in localized areas, as well as more widespread bora and chinook winds in the winter season (October–March). The predictors and general weather patterns that result in strong winds across the region are well understood by local forecasters. However, no single predictor provides notable skill by itself in separating warning-level events from others. Random forest (RF) classifier models were developed to improve upon high wind prediction using a training dataset constructed of archived observations and model parameters from the North American Regional Reanalysis (NARR). Three locations were selected for initial RF model development, including the city of Cheyenne, Wyoming, and two gap regions along Interstate 80 (Arlington) and Interstate 25 (Bordeaux). Verification scores over two winters suggested the RF models were beneficial relative to current operational tools when predicting warning-criteria high wind events. Three case studies of high wind events provide examples of the RF models’ effectiveness to forecast operations over current forecast tools. The first case explores a classic, widespread high wind scenario, which was well anticipated by local forecasters. A more marginal scenario is explored in the second case, which presented greater forecast challenges relating to timing and intensity of the strongest winds. The final case study carefully uses Global Forecast System (GFS) data as input into the RF models, further supporting real-time implementation into forecast operations.

THERMOBAROGEOCHEMICAL FUNDAMENTALS AND APPLIED POSSIBILITIES OF LOCAL FORECAST, RESEARCH AND ASSESSMENT OF POST-MAGMATIC MINERALIZATION

Important issues of local forecasting, search and assessment of post-magmatic mineralization during various stages of geological exploration are analysed on a thermobarogeochemical (TBGC) basis. The theoretical basis of TBGC-forecasting, exploration and assessment of mineralization is the stability of the regime of physical-chemical conditions for the formation of productive mineral paragenesis. They are formed in a fairly narrow range of TBGC-parameters changes of the specific to chemical composition and aggregate-density state of ore-forming medium at certain (optimal) temperature and pressure values. This is revealed in the phase typomorphism of the respective families of fluid inclusions practically regardless of geotectonic conditions and metallogenic specialization of the ore regions. For example, the molybdenum-tungsten formation of the greisen type is characterized by the following: alkaline-halogen (fluoride-chloride-potassium-sodium) composition of inclusions with a high salt concentration (65-35 wt.% NaCl), and the spread of syngenetic families of inclusions with minerals-"prisoners" (which testify to the boiling processes of solutions) and solutions of inclusions of critical density. On the other hand, in gold ore formations, the productive gold-bearing stages in the corresponding paragenesises contain water-carbon dioxide inclusions, which homogenize at 290-180°C under conditions of intensive heterogenization and degassing (СО2) of solutions with extremely different phase ratios (G-LСО2-Н2О; LСО2-LН2О; G-LСО2; LСО2-L; LСО2-GСО2). Fluid inclusions of two- and one-phase CO2 with wide variations in density (from 1.02 to 0.4-0.1 g/cm3 for various depth formations) and simultaneous homogenization into liquid and gas phases (boiling) are common. So, we determine the distribution of molybdenum-tungsten or gold mineralization by the distribution of the corresponding families of syngenetic inclusions. An equally important prerequisite for the implementation of applied TBGC problems is the possibility of diagnosis and spatial extrapolation of these parameters (TBGC-zonation) with the determination of the spatial position of zones that are physically and chemically favourable for the development of post-magmatic mineralization. Such tools can be the vertical paleotemperature gradient ∆Т/100 m and its inverse functional relationship with the vertical spread of mineralization, the ratio of the amount of reduced gases to the amount of oxidized gases, etc. Specific examples of mineralization forecasting, search, and assessment using the genetic features of the deposit and the complex application of TBGC-criteria for spatial-temporal modelling of mineralization distribution are given.

Machine Learning-assisted spatiotemporal chaos forecasting

Long-term forecasting of extreme events such as oceanic rogue waves, heat waves, floods, earthquakes, has always been a challenge due to their highly complex dynamics. Recently, machine learning methods have been used for model-free forecasting of physical systems. In this work, we investigated the ability of these methods to forecast the emergence of extreme events in a spatiotemporal chaotic passive ring cavity by detecting the precursors of high intensity pulses. To this end, we have implemented supervised sequence (precursors) to sequence (pulses) machine learning algorithms, corresponding to a local forecasting of when and where extreme events will appear.

Organizational Modes of Spring and Summer Convective Storms and Associated Severe Weather over Southern China during 2015–19

Abstract This study investigates the organizational modes of convective storms and associated severe weather in spring and summer (March–August) of 2015–19 over southern China. These storms are classified into three major organizational structures (cellular, linear, and nonlinear), including 10 dominant morphologies. In general, cellular systems are most frequent, followed by linear systems. Convective storms are common in spring, increasing markedly from April to June, and peak in June. Convective storm cases are usually longer lived in spring, while shorter lived in summer. They also present pronounced diurnal variations, with a primary peak in the afternoon and several secondary peaks during the night to the morning. Approximately 79.7% of initial convection clearly exhibits a dominant eastward movement, with a faster moving speed in spring. Convective storms frequently evolve among organizational modes during their life spans. Linear systems produce the most severe weather observations, in which convective lines with trailing stratiform rain are most prolific. Bow echoes are most efficient in producing severe weather events among all systems, despite their rare occurrences. In spring, lines with parallel stratiform rain are abundant producers of severe wind events, ranking the second highest probability. In summer, embedded lines produce the second largest proportion of intense rainfall events, whereas lines with leading stratiform rain are most efficient in generating extremely intense rainfall and thus pose a distinct flooding threat. Broken lines produce the largest proportion of severe weather events among cellular storms. In contrast, nonlinear systems possess the weakest capability to produce severe weather events. Significance Statement Under the influence of the East Asian summer monsoon, severe weather events produced by convective storms occur frequently in China, leading to serious natural disasters. Numerous studies have demonstrated that the morphologies of convective storms are helpful to improve our understanding and prediction of convective storms. However, fewer attempts have been made to examine the convective morphologies over southern China. We aim to reveal the general features of convective organizational modes (e.g., frequencies, durations, variations, etc.) and determine which particular types of severe weather are more or less likely to be associated with particular convective morphologies. These results are of benefit to local forecasters for better anticipating the storm types and issuing warnings for related hazardous weather.

Data-Driven Strategies for Hierarchical Predictive Control in Unknown Environments

This article proposes a hierarchical learning architecture for safe data-driven control in unknown environments. We consider a constrained nonlinear dynamical system and assume the availability of state-input trajectories solving control tasks in different environments. In addition to task-invariant system state and input constraints, a parameterized environment model generates task-specific state constraints, which are satisfied by the stored trajectories. Our goal is to use these trajectories to find a safe and high-performing policy for a new task in a new, unknown environment. We propose using the stored data to learn generalizable control strategies. At each time step, based on a local forecast of the new task environment, the learned strategy consists of a target region in the state space and input constraints to guide the system evolution to the target region. These target regions are used as terminal sets by a low-level model predictive controller. We show how to <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">i)</i> design the target sets from past data and then <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ii)</i> incorporate them into a model predictive control scheme with shifting horizon that ensures safety of the closed-loop system when performing the new task. We prove the feasibility of the resulting control policy, and apply the proposed method to robotic path planning, racing, and computer game applications. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —This paper was motivated by the challenge of designing safe controllers for autonomous systems navigating through new environments. We consider scenarios where trajectory data from control tasks in different environments is available to the control designer. Possible applications include autonomous vehicles racing on new tracks or robotic manipulators performing tasks in the presence of new obstacles. Existing approaches to model-based control design for new environments generally use trajectory libraries, systematically adapting stored trajectories to the constraints of the new environment. This typically requires a priori knowledge of the entire task environment as well as resources to store and maintain the growing library. This paper suggests a new hierarchical control approach, in which stored trajectories are used to learn high-level strategies that can be applied while solving the new task. The strategies are learned offline, and only the parameterized strategy function needs to be stored for online control. Strategies only require knowledge of the nearby task environment, and provide navigation guidelines for the system. In this paper we show how to find such strategies from previous task data and how to integrate them into a low-level controller to safely and efficiently solve the new task. We also show how to adapt the modular framework as needed for a user’s desired application. Simulation experiments in robotic manipulator, autonomous vehicle, and computer game examples suggest that our approach can be used in a wide range of applications. In future research, we will address how to adapt the method for time-varying or stochastic environments.

Diagnostic Analysis of Multimodel Rainstorm Forecast for Cases Based on MODE Method

It is very important to analyze the performance of the model rainstorm forecast for understanding the intensity and position deviation of the model precipitation and improving the forecast ability. This paper uses classical scoring and the MODE (Method for Object-Based Diagnostic Evaluation) method to evaluate the forecast performance of different products. The forecast and observation data used in the article mainly include CMA (China Meteorological Administration) multi-source merged precipitation analysis, the precipitation forecast of ECMWF (European Centre for Medium-Range Weather Forecasts), CMA-Meso (Mesoscale model forecast of CMA) and SCMOC (National Meteorological Center grid precipitation forecast guidance product) data. At the same time, the possible correction method of heavy rainfall area is explored by using the high and low-level circulation configuration of the ECMWF model. The main conclusions are as follows: ① MODE spatial verification shows that the number, intensity, area and location of ECMWF rainstorm precipitation objects match the observed precipitation best, which is obviously better than SCMOC and CMA-Meso precipitation forecasts. There are significantly fewer SCMOC rainstorm precipitation objects, and the area of each single precipitation object is significantly larger, which often fails to report the small area objects of convective precipitation. ② The reason for the high TS score of SCMOC is that it reduces the number of small area rainstorm objects and avoids the “double punishment” phenomenon caused by position forecast error, which leads to the failure of SCMOC in local rainstorm forecasts. ③ Analyzing the relationship between the circulation situation of the ECMWF model and the location of rainstorm forecasts by the model, it is found that the location of the rainstorm area is consistent with the upper circulation system, especially with the strong rising area of the vertical velocity of 700 hPa and the high value area of the specific humidity of 850 hPa. When the rainstorm area coincides with the upper air system but is not consistent with the ground convergence area and the high value area of velocity potential, the rainstorm location often has a large deviation. The location of the surface convergence area can be used as a reference to improve the performance of the rainstorm forecast.

Capacity Building Needs of Artisanal Fisherfolks in North Central, Nigeria

The study analyzed the capacity building needs of artisanal fisherfolks in North Central, Nigeria. The Data were collected with the use of a questionnaire and analysed using frequency, percentages, mean score and Correlation. The results showed that about 99.2% of the fishers were male and the mean age of the fishers was 43.9ears. The mean years of experience were 24.3. The mean quantity of catch per day was 17.9 Kg and 6 persons was the average household size. About 47.33 % of the respondents has a high knowledge level on improved fishing practices. The fisherfolks required capacity building in handling and transportation of captured fish (x ̅=6.59), use and repair of nets, cages and traps (mean=6.29), use and maintenance of multiple hooks on a line (x ̅=6.71), canoe use and maintenance (x ̅=6.74), use of chorkor smoker (x ̅=6.68), fish marketing and distribution (x ̅=6.65), credit acquisition (x ̅=5.92), maintaining personal and environmental health hygiene (x ̅=6.0) and local weather forecasting (x ̅=6.32). The highest ranked constraints affecting artisanal fishing practices were the high cost of fishing inputs and inadequate access to credit facilities (x ̅=2.72). The study therefore recommends the provision of training in the identified areas of capacity building needs and the provision of more frequent and effective extension services to the fisherfolks.

Hourly solar irradiation forecast using hybrid local gravitational clustering and group method of data handling methods.

The foundation for many solar energy uses as well as economic and environmental concerns is global solar irradiation information. However, due to solar irradiation and measurement variations, reliable worldwide statistics on solar irradiation are frequently impossible or challenging to acquire. In addition, more precise forecasts of solar irradiation play an increasingly important role in electric energy planning and management due to integrating photovoltaic solar systems into power networks. Hence, this paper proposes a new hybrid model for 1-h ahead solar irradiation forecasting called LGC-GMDH (local gravitational clustering-group method of data handling). The novel LGC-GMDH model is based on local clustering that adequately captures the underlying features of the solar irradiation time series. Each cluster is then forecasted using the GMDH method, which is a self-organized system capable of handling very complicated nonlinear problems. Finally, these local forecasts are reconstructed in order to obtain the global forecast. Comparative study between the proposed model and the traditional individual models such as backpropagation neural network (BP), supporting vector machines (SVM), long short-term memory (LTSM), and hybrid models such as BP-MLP, RNN-MLP, LSTM-MLP hybrid wavelet packet decomposition (WPD), convolutional neural network (CNN) with LSTM-MLP, and ANFIS clustering shows that the proposed model overcomes conventional model deficiencies and achieves more precise predicting outcome.

Development of model output statistics based on the least absolute shrinkage and selection operator regression for forecasting next‐day maximum temperature in South Korea

AbstractRegression models for model output statistics (MOS) based on least absolute shrinkage and selection operator methods were developed to forecast next‐day maximum surface air temperature (TMAX) during the warm season in South Korea. The forecast fields from the operational numerical weather prediction (NWP) system of the Korean Meteorological Administration for global and local forecasts and the observed TMAX data in 225 observation stations were used as input variables for the MOS. The training period was July and August (JA) from 2015 to 2018, and the regression models were tested using data from JA 2019. As a result of hindcasting for the test period, the MOS models performed significantly better for next‐day TMAX forecasting over South Korea than the numerical models during JA 2019. The mean TMAX errors were reduced by over 1°C in MOSs compared to those in the numerical models. However, the TMAX forecast performance was generally lower in the higher‐resolution NWP Local Data Assimilation and Prediction System (LDAPS)‐based MOS (LMOS) than in the lower‐resolution NWP Global Data Assimilation and Prediction System (GDAPS)‐based MOS. This pattern was dominant when LDAPS simulated the TMAX more accurately than average. In particular, the random TMAX error of LDAPS was larger than that of GDAPS during the training period, and a positive random error of TMAX was magnified in LMOS. Because the other predictors forecasted from LDAPS can be associated with lower TMAX forecast performance of LMOS, in addition to TMAX effects as a predictor, a new MOS was developed using both LDAPS and GDAPS outputs. The forecast accuracy was improved by up to 0.3°C when the forecast fields from the GDAPS substituted several LMOS predictors, even though TMAX was the primary predictor for LMOS.