Weather Prediction Methods Research Articles

A Data-Driven Method for Hybrid Data Assimilation with Multilayer Perceptron

Abstract Accurate and timely weather prediction is of significance for autonomous vehicles, such as designing more appropriate sensors or other configurations and developing safer driving strategies. Generally, as the mainstream weather prediction method, numerical weather prediction (NWP) relies on high-quality spatio-temporal observations. However, the precise state of the real world is not measurable. Thus, how to obtain a proper initial condition estimation based on big geospatial-temporal data is a crucial procedure for NWP. Data assimilation (DA) has been a traditional solution to the problem, for the better performance of which various mathematical-physics models have been used. However, the computational effectiveness and efficiency are still largely compromised by the complicated and nonparallel integration process in existing DA methods. In this paper, we propose a novel data-driven method named HDA-MLP to address the DA problem. We first constructed a customized MLP by introducing the temporal peculiarities of the state variables to simulate and optimize pure 3DVar and EnKF. Then we blended the optimized analysis fields directly by implicitly updating the background error covariance matrix through another neural network model to alleviate the dependence on traditional DA methods. We conducted extensive experiments to investigate the effectiveness and efficiency of the proposal by utilizing two classical nonlinear dynamic models. Results reveal that our approach has better robustness and enhanced capability to capture the variation of state variables. Notably, the analysis quality and computational efficiency are significantly improved.

Deep Learning-Based Weather Prediction: A Survey

Abstract Weather forecasting plays a fundamental role in the early warning of weather impacts on various aspects of human livelihood. For instance, weather forecasting provides decision making support for autonomous vehicles to reduce traffic accidents and congestions, which completely depend on the sensing and predicting of external environmental factors such as rainfall, air visibility and so on. Accurate and timely weather prediction has always been the goal of meteorological scientists. However, the conventional theory-driven numerical weather prediction (NWP) methods face many challenges, such as incomplete understanding of physical mechanisms, difficulties in obtaining useful knowledge from the deluge of observation data, and the requirement of powerful computing resources. With the successful application of data-driven deep learning method in various fields, such as computer vision, speech recognition, and time series prediction, it has been proven that deep learning method can effectively mine the temporal and spatial features from the spatio-temporal data. Meteorological data is a typical big geospatial data. Deep learning-based weather prediction (DLWP) is expected to be a strong supplement to the conventional method. At present, many researchers have tried to introduce data-driven deep learning into weather forecasting, and have achieved some preliminary results. In this paper, we survey the state-of-the-art studies of deep learning-based weather forecasting, in the aspects of the design of neural network (NN) architectures, spatial and temporal scales, as well as the datasets and benchmarks. Then we analyze the advantages and disadvantages of DLWP by comparing it with the conventional NWP, and summarize the potential future research topics of DLWP.

Finite Difference Methods for Weather Prediction in Abuja, Nigeria

In this research, we have been able to simulate some finite difference schemes to predict weather trends of Abuja Station, Nigeria. By analyzing the results from these schemes, it has shown that the best scheme in the finite difference method that gives a close accurate weather forecast is the trapezoidal scheme hence we use it to simulate numerical weather data obtained from Federal Airports Authority of Nigeria (FAAN), Abuja and corresponding numerical weather data obtained by the compatible finite difference schemes, using MATLAB (R2012a) software to obtain future numerical weather trends.

Solar irradiance resource and forecasting: a comprehensive review

With the increase in demand for energy, penetration of alternative sources of energy in the power grid has increased. Photovoltaic (PV) energy is the most common and popular form of energy sources which is widely integrated into the existing grid. As solar energy is intermittent in nature, to ensure uninterrupted and reliable power supply to the prosumers, it is essential to forecast the solar irradiance. Accurate solar forecasting is necessary to facilitate large-scale modelling and deployment of PV plants without disrupting the quality and reliability of the power grid as well as to manage the power demand and supply. There are various methods to predict the solar irradiance such as numerical weather prediction methods, satellite-based approaches, cloud-image based methodologies, data-driven methods, and sensor-network based approaches. This study gives an overall review of the different resources and methods used for forecasting solar irradiance in different time horizons and also gives an extensive review of the sensor networks that are used for determining solar irradiance. The various error metrics and accessible data sets available for the sensor networks are also discussed that can be used for validation purposes.

A Comprehensive Wind Power Forecasting System Integrating Artificial Intelligence and Numerical Weather Prediction

The National Center for Atmospheric Research (NCAR) recently updated the comprehensive wind power forecasting system in collaboration with Xcel Energy addressing users’ needs and requirements by enhancing and expanding integration between numerical weather prediction and machine-learning methods. While the original system was designed with the primary focus on day-ahead power prediction in support of power trading, the enhanced system provides short-term forecasting for unit commitment and economic dispatch, uncertainty quantification in wind speed prediction with probabilistic forecasting, and prediction of extreme events such as icing. Furthermore, the empirical power conversion machine-learning algorithms now use a quantile approach to data quality control that has improved the accuracy of the methods. Forecast uncertainty is quantified using an analog ensemble approach. Two methods of providing short-range ramp forecasts are blended: the variational doppler radar analysis system and an observation-based expert system. Extreme events, specifically changes in wind power due to high winds and icing, are now forecasted by combining numerical weather prediction and a fuzzy logic artificial intelligence system. These systems and their recent advances are described and assessed.

Time series forecasting on multivariate solar radiation data using deep learning (LSTM)

Energy management is an emerging problem nowadays and utilization of renewable energy sources is an efficient solution. Solar radiation is an important source for electricity generation. For effective utilization, it is important to know precisely the amount from different sources and at different horizons: minutes, hours, and days. Depending on the horizon, two main classes of methods can be used to forecast the solar radiation: statistical time series forecasting methods for short to midterm horizons and numerical weather prediction methods for medium- to long-term horizons. Although statistical time series forecasting methods are utilized in the literature, there are a limited number of studies that utilize deep artificial neural networks. In this study, we focus on statistical time series forecasting methods for short-term horizons (1 h). The aim of this study is to discover the effect of using multivariate data on solar radiation forecasting using a deep learning approach. In this context, we propose a multivariate forecast model that uses a combination of different meteorological variables, such as temperature, humidity, and nebulosity. In the proposed model, recurrent neural network (RNN) variation, namely a long short-term memory (LSTM) unit is used. With an experimental approach, the effect of each meteorological variable is investigated. By hyperparameter tuning, optimal parameters are found in order to construct the best models that fit the global solar radiation data. We compared the results with those of previous studies and we found that the multivariate approach performed better than the previous univariate models did. In further experiments, the effect of combining the most effective parameters was investigated and, as a result, we observed that temperature and nebulosity are the most effective parameters for predicting future solar radiance.

Research Progress of Interplanetary Physics in Mainland China

Significant progress has been made by Chinese scientists in research of interplanetary physics during the recent two years (2018-2020). These achievements are reflected at least in the following aspects:Activities in solar corona and lower solar atmosphere; solar wind and turbulence; filament/prominence, jets, flares, and radio bursts; active regions and solar eruptions; coronal mass ejections and their interplanetary counterparts; other interplanetary structures; space weather prediction methods; magnetic reconnection; Magnetohydrodynamic (MHD) numerical modeling; solar energetic particles, cosmic rays, and Forbush decreases; machine learning methods in space weather and other aspects. More than one hundred and forty papers in the academic journals have been published in these research directions. These fruitful achievements are obtained by Chinese scholars in solar physics and space physics either independently or through international collaborations. They greatly improve people's understanding of solar activities, solar eruptions, the corresponding space weather effects, and the Sun-Earth relations. Here we will give a very brief review on the research progress. However, it must be pointed out that this paper may not completely cover all achievements in this field due to our limited knowledge.

Scoping Low-Cost Measures to Nowcast Sub-Hourly Solar Radiations for Buildings

Solar energy, as a clear, renewable and promising resource, has brought tremendous value for carbon emission reduction in the last three decades. However, the uncertainty of solar radiation brings significant challenges to the renewable heat and electrical system for buildings. Thus, many different solar radiation forecasting methods, such as Numerical Weather Prediction Methods (NWP), Statistical Methods, Top-Down Methods, Bottom-Up Methods and Hybrid Methods have been developed to make the best guess on future solar radiations. Based on the meteorology definition, nowcasting refers to the forecast of the temporal horizon from the next few seconds up to six hours. Predicting solar radiation within this range are extremely challenging, but crucial for solar energy system operation. This paper firstly reviewed the state of art of solar radiation forecasting methods and compared the key features of each method. According to the advantage and costs of each method, this paper then proposed a methodology to generate high accuracy low-cost solar radiation nowcast data for optimising building solar thermal system performance and related control strategies. Note that this paper focused on the scoping study of methodology rather than field experiments, and further research with experiments data will be reported in a journal publication.

Solar energy production forecasting through artificial neuronal networks, considering the Föhn, north and south winds in San Juan, Argentina

This study presents a method to predict a day-ahead solar irradiation curve, under extreme meteorological phenomena (Föhn, north and south winds), existing in the province of San Juan, Argentina. The proposed method is based on an artificial neuronal network (ANN) which is trained with a data set filtered by the environmental variables that characterise the mentioned phenomena. A previously calculated ideal solar irradiation curve is modified from the forecasts generated by the ANN. The proposed methodology merges statistical learning methods and numerical weather prediction (NWP) methods, typically used to improve upon the raw forecast of a NWP model. A reduction of the uncertainty in the power production of photovoltaic plants in San Juan can be achieved with the results of the proposed forecasting method.

IMPLEMENTATION OF ARTIFICIAL NEURAL NETWORK FOR MONTHLY RAIN ATTACKS IN SOUTH KALIMANTAN

Prakiraan hujan bulanan bisa digunakan untuk antisipasi banjir dan manajemen sumber daya air, keselamatan jiwa dan harta benda, serta keberlangsungan aktivitas ekonomi. Penggunaan Jaringan Syaraf Tiruan sebagai bagian dari Machine Learning adalah teknik yang sering digunakan selain numerical weather prediction dan metode statistik. Menggunakan peru-bah data bulan dan data empirical orthogonal function anomali suhu muka laut bulanan pada 12 lokasi menghasilkan nilai korelasi yang baik saat pembuatan model, tetapi hasil verifikasi menunjukkan akurasi yang baik didapatkan saat periode musim kemarau dan skill terjelek saat peralihan musim kemarau ke musim hujan.

Daily Precipitation Fields Modeling across the Great Lakes Region (Canada) by Using the CFSR Reanalysis

AbstractReanalyses, generated by numerical weather prediction methods assimilating past observations, provide consistent and continuous meteorological fields for a specific period. In regard to precipitation, reanalyses cannot be used as a climate proxy of the observed precipitation, as biases and scale mismatches exist between the datasets. In the present study, a stochastic model output statistics (SMOS) approach combined with meta-Gaussian spatiotemporal random fields was employed to cope with these caveats. The SMOS is based on the generalized linear model (GLM) and the vector generalized linear model (VGLM) frameworks to model the precipitation occurrence and intensity, respectively. Both models use the Climate Forecast System Reanalysis (CFSR) precipitation as covariate and were locally calibrated at 173 sites across the Great Lakes region. Combined with meta-Gaussian random fields, the GLM and VGLM models allowed for the generation of spatially coherent daily precipitation fields across the region. The results indicated that the approach corrected systematic biases and provided an accurate spatiotemporal structure of daily precipitation. Performances of selected precipitation indicators from the joint Commission for Climatology (CCl)/CLIVAR/JCOMM Expert Team on Climate Change Detection and Indices (ETCCDI) were good and were systematically improved when compared to CFSR.

Stochastic Post-Processing of CFSR Daily Precipitation across Canada

ABSTRACTReanalyses, based on numerical weather prediction methods assimilating past observations, provide continuous precipitation datasets and represent interesting options for assessing the climatology of regions with sparse station networks (e.g., northern Canada). However, reanalysis series cannot be used directly because of possible biases and mismatch between their spatial and temporal resolutions with that needed for local applications. To address these issues, a Stochastic Model Output Statistics (SMOS) approach was selected to post-process precipitation series simulated by the Climate Forecast System Reanalysis (CFSR) across Canada. This approach uses CFSR precipitation as a covariate and is based on two regression models: the first one is a logistic regression that deals with precipitation occurrence, and the second is a vector generalized linear model for precipitation intensity. At-site post-processed daily precipitation series are randomly generated using the SMOS approach, and selected climate indicators from the Expert Team on Climate Change Detection and Indices, which is jointly sponsored by the Commission for Climatology of the World Meteorological Organization's (WMO) World Climate Data and Monitoring Programme, the Climate Variability and Predictability Programme of the World Climate Research Programme, and the Joint WMO-IOC Technical Commission for Oceanography and Marine Meteorology (CCI/CLIVAR/JCOMM) are estimated and compared with corresponding observed and CFSR values. The two models in the SMOS approach, in addition to adequately correcting systematic biases, produced better predictions than the climatology of the wet and dry and intensity sequences. Additionally, the SMOS generally yields consistent climate indices when compared with those from CFSR without post-processing, though there is still room for improvement for specific indices (e.g., annual maximum of cumulative wet days).

딥러닝 기법을 이용한 내일강수 예측

Abstract For accurate precipitation forecasts the choice of weather factors and prediction method is very important. Recently, machine learning has been widely used for forecasting precipitation, and artificial neural network, one of machine learning techniques, showed good performance. In this paper, we suggest a new method for forecasting precipitation using DBN, one of deep learning techniques. DBN has an advantage that initial weights are set by unsupervised learning, so this compensates for the defects of artificial neural networks. We used past precipitation, temperature, and the parameters of the sun and moon's motion as features for forecasting precipitation. The dataset consists of observation data which had been measured for 40 years from AWS in Seoul. Experiments were based on 8-fold cross validation. As a result of estimation, we got probabilities of test dataset, so threshold was used for the decision of precipitation. CSI and Bias were used for indicating the precision of precipitation. Our experimental results showed that DBN performed better than MLP. Key Words : Deep learning, Deep belief network, Precipitation, Forecast본 논문은 예보기술지원 및 활용 연구(NIMR-2012-B-1)의 일환으로 수행되었습니다.This is an Open-Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Communication and interpretation of regional weather forecasts: a survey of the Italian public

The aim of the present study is to contribute to the correctness and effectiveness of weather forecast communication, the importance of which has been steadily growing along with the improvement in numerical weather prediction models and methods as well as the general awareness about the increase of extreme events within a context of global climate change. An extensive survey was conducted among the general users of the weather forecasts issued by the regional meteorological service of Tuscany, Italy (LaMMA Consortium), which resulted in 2388 volunteers responding to the questions aimed at better understanding of how people access, interpret and use weather forecasts. The survey also includes some items investigated in previous research, allowing comparison with similar findings in other countries. The most critical issue concerns the uncertainty information, investigated with the main aim of verifying the existence and relevance of inferential mechanisms in the interpretation of weather icons and maps used in LaMMA forecasts to assess uncertainty. The present study also discusses users' interpretations of the probability of precipitation forecasts and their preferences on how forecast uncertainty is conveyed. Results show that, even if the Italian public is accustomed to strictly deterministic weather forecasts, people attribute uncertainty to them on their own even if lacking any explicit indication, thus suggesting the need to supplement the existing forecasts with both graphical and textual information about uncertainty, particularly in the case of precipitation forecasts.

Multi-model ensemble: technique and validation

Abstract. In this study, a method of numerical weather prediction by ensemble for the South American region is proposed. This method takes into account combinations of the numerical predictions of various models, assigning greater weight to models that exhibit the best performance. Nine operational numerical models were used to perform this study. The main objective of the study is to obtain a weather forecasting product (short-to-medium range) that combines what is best in each of the nine models used in the study, thus producing more reliable predictions. The proposed method was evaluated during austral summer (December 2012, and January and February 2013) and winter (June, July and August 2013). The results show that the proposed method can significantly improve the results provided by the numerical models and consequently has promising potential for operational applications in any weather forecasting center.

Knowing Climate Change: Local Social Institutions and Adaptation in Indian Groundwater Irrigation

How do farmers understand and predict weather variability; in what ways are local weather prediction techniques institutionalized (e.g., in new agricultural practices or cropping decisions) to meet the agricultural challenges of climate change–induced socioecological variability; and how do these vary across space, according to socioecological difference? This article examines these questions through a case study examination of local weather prediction methods and adaptive strategies to ongoing weather-related variability by groundwater-dependent, irrigating farmers in Rajasthan, India. Conducted in 2009 and 2011, the work finds, first, that farmers rely on multiple local methods of weather prediction, which, along with multiple and often conflicting social and ecological factors, inform their cropping decisions. Second, these prediction methods and associated cropping strategies interact with a number of strategies to mitigate weather and agrarian variability more generally, such as new cropping strategies, seasonal migration, and market articulations. The article advances our thinking about what climate change, as yet another (but perhaps unique) agrarian perturbation, means for farmers’ livelihoods. It concludes with a discussion of the implications of the analysis for formal climate change adaptation strategies and related ongoing groundwater policy.

A comparison of two numerical weather prediction methods for diagnosing fast‐physics errors in climate models

AbstractAn important aspect of the assessment of climate models is the representation of fast physical processes, which have a strong impact on climate projections. Since these fast processes also have a strong impact on weather forecasts, there is potential in using numerical weather prediction (NWP)‐based methodologies in climate model assessments.Two NWP‐based methods have been proposed in the literature. The first is referred to as the ‘initial tendency’ approach and assesses the systematic departure in the first few hours of short climate model resolution forecasts from the weather trajectory. For this method, the forecast model needs data assimilation capabilities to create initial conditions consistent with the model physics. The second approach is referred to as the ‘transpose‐AMIP’ method. This makes use of analyses produced with a different forecast model to initiate weather forecasts with a climate model. It is generally recommended that errors are diagnosed after a few forecast days to minimise the impact of the non‐native initial conditions.Here we compare these two methods in one NWP model over the subtropical South Pacific, which encompasses climatologically important regimes and transitions between them. The impact of the non‐native initial conditions is compared with the impact of an error introduced in the convection scheme.Results show that care is needed in the design and evaluation of transpose‐AMIP‐style experiments if model errors are to be disentangled from, at short lead‐times, the initial shock of using non‐native initial conditions and, at longer lead‐times, the complicating effects of interactions and the growth of chaos. On the other hand, the initial tendency approach is able to identify the introduced model error. This provides some evidence that it might be worthwhile for climate centres to explore the use of data assimilation.

Research Progress of Solar Corona and Interplanetary Physics in China: 2010-2012

The scientific objective of solar corona and interplanetary research is the understanding of the various phenomena related to solar activities and their effects on the space environments of the Earth. Great progress has been made in the study of solar corona and interplanetary physics by the Chinese space physics community during the past years. This paper will give a brief report about the latest progress of the corona and interplanetary research in China during the years of 2010?2012. The paper can be divided into the following parts: solar corona and solar wind, CMEICME, magnetic reconnection, energetic particles, space plasma, space weather numerical modeling by 3D SIP-CESE MHD model, space weather prediction methods, and proposed missions. They constitute the abundant content of study for the complicated phenomena that originate from the solar corona, propagate in interplanetary space, and produce geomagnetic disturbances. All these progresses are acquired by the Chinese space physicists, either independently or through international collaborations.

Regression modeling method of space weather prediction

A regression modeling method of space weather prediction is proposed. It allows forecasting Dst index up to 6 hours ahead with about 90% correlation. It can also be used for constructing phenomenological models of interaction between the solar wind and the magnetosphere. With its help two new geoeffective parameters were found: latitudinal and longitudinal flow angles of the solar wind. It was shown that Dst index remembers its previous values for 2000 hours.

Inter-Comparison Study of the ENSEMBLE Project

During the days of the Chernobyl accident, the European national long-range dispersion forecasts would differ because of differences in national models, and differences in weather prediction methods. ENSEMBLE project was launched for a reconciliation and harmonization of the disparate long-range dispersion forecasts. Responsible European emergency organizations in addition to Canadian, Japanese, Korean and US agencies have participated in ENSEMBLE. KAERI joined the inter-comparison study for the exercise on the 901–001 scenario in ENSEMBLE. KAERI was assigned KR1 as a national code and 53 as a model number. The model of KAERI was compared with the other models of the participants in ENSEMBLE. The comparative results are presented with the scatter plots and statistical methods in this paper.