AI-based Correction of Wave Forecasts Using the Transformer-enhanced UNet Model

Grid forecasting can be used to effectively enhance the spatial and temporal density of forecast products,thereby improving the capability of short-term marine disaster forecasting and warnings in terms of proximity.The traditional method that relies on forecasters'subjective correction of station observation data for forecasting has been unable to meet the practical needs of refined forecasting.To address this problem,this paper proposes a Transformer-enhanced UNet(TransUNet)model for wave forecast AI correction,which fuses wind and wave information.The Transformer structure is integrated into the encoder of the UNet model,and instead of using the traditional upsampling method,the dual-sampling module is employed in the decoder to enhance the feature extraction capability.This paper compares the TransUNet model with the traditional UNet model using wind speed forecast data,wave height forecast data,and significant wave height reanalysis data provided by ECMWF.The experimental results indicate that the TransUNet model yields smaller root-meansquare errors,mean errors,and standard deviations of the corrected results for the next 24-h forecasts than does the UNet model.Specifically,the root-mean-square error decreased by more than 21.55%compared to its precorrection value.According to the statistical analysis,87.81%of the corrected wave height errors for the next 24-h forecast were within±0.2m,with only 4.56%falling beyond±0.3 m.This model effectively limits the error range and enhances the ability to forecast wave heights.

Decadal Forecasts of Large Earthquakes along the Northern San Andreas Fault System, California: Increased Activity on Regional Creeping Faults Prior to Major and Great Events

The three largest earthquakes in northern California since 1849 were preceded by increased decadal activity for moderate-size shocks along surrounding nearby faults. Increased seismicity, double-difference precise locations of earthquakes since 1968, geodetic data and fault offsets for the 1906 great shock are used to re-examine the timing and locations of possible future large earthquakes. The physical mechanisms of regional faults like the Calaveras, Hayward and Sargent, which exhibit creep, differ from those of the northern San Andreas, which is currently locked and is not creeping. Much decadal forerunning activity occurred on creeping faults. Moderate-size earthquakes along those faults became more frequent as stresses in the region increased in the latter part of the cycle of stress restoration for major and great earthquakes along the San Andreas. They may be useful for decadal forecasts. Yearly to decadal forecasts, however, are based on only a few major to great events. Activity along closer faults like that in the two years prior to the 1989 Loma Prieta shock needs to be examined for possible yearly forerunning changes to large plate boundary earthquakes. Geodetic observations are needed to focus on identifying creeping faults close to the San Andreas. The distribution of moderate-size earthquakes increased significantly since 1990 along the Hayward fault but not adjacent to the San Andreas fault to the south of San Francisco compared to what took place in the decades prior to the three major historic earthquakes in the region. It is now clear from a re-examination of the 1989 mainshock that the increased level of moderate-size shocks in the one to two preceding decades occurred on nearby East Bay faults. Double-difference locations of small earthquakes provide structural information about faults in the region, especially their depths. The northern San Andreas fault is divided into several strongly coupled segments based on differences in seismicity.

Non-crossing Quantile Regression Neural Network as a Calibration Tool for Ensemble Weather Forecasts

Despite the maturity of ensemble numerical weather prediction(NWP),the resulting forecasts are still,more often than not,under-dispersed.As such,forecast calibration tools have become popular.Among those tools,quantile regression(QR)is highly competitive in terms of both flexibility and predictive performance.Nevertheless,a long-standing problem of QR is quantile crossing,which greatly limits the interpretability of QR-calibrated forecasts.On this point,this study proposes a non-crossing quantile regression neural network(NCQRNN),for calibrating ensemble NWP forecasts into a set of reliable quantile forecasts without crossing.The overarching design principle of NCQRNN is to add on top of the conventional QRNN structure another hidden layer,which imposes a non-decreasing mapping between the combined output from nodes of the last hidden layer to the nodes of the output layer,through a triangular weight matrix with positive entries.The empirical part of the work considers a solar irradiance case study,in which four years of ensemble irradiance forecasts at seven locations,issued by the European Centre for Medium-Range Weather Forecasts,are calibrated via NCQRNN,as well as via an eclectic mix of benchmarking models,ranging from the naïve climatology to the state-of-the-art deep-learning and other non-crossing models.Formal and stringent forecast verification suggests that the forecasts post-processed via NCQRNN attain the maximum sharpness subject to calibration,amongst all competitors.Furthermore,the proposed conception to resolve quantile crossing is remarkably simple yet general,and thus has broad applicability as it can be integrated with many shallow-and deep-learning-based neural networks.

Downscaling Seasonal Precipitation Forecasts over East Africa with Deep Convolutional Neural Networks

This study assesses the suitability of convolutional neural networks(CNNs) for downscaling precipitation over East Africa in the context of seasonal forecasting. To achieve this, we design a set of experiments that compare different CNN configurations and deployed the best-performing architecture to downscale one-month lead seasonal forecasts of June–July–August–September(JJAS) precipitation from the Nanjing University of Information Science and Technology Climate Forecast System version 1.0(NUIST-CFS1.0) for 1982–2020. We also perform hyper-parameter optimization and introduce predictors over a larger area to include information about the main large-scale circulations that drive precipitation over the East Africa region, which improves the downscaling results. Finally, we validate the raw model and downscaled forecasts in terms of both deterministic and probabilistic verification metrics, as well as their ability to reproduce the observed precipitation extreme and spell indicator indices. The results show that the CNN-based downscaling consistently improves the raw model forecasts, with lower bias and more accurate representations of the observed mean and extreme precipitation spatial patterns. Besides, CNN-based downscaling yields a much more accurate forecast of extreme and spell indicators and reduces the significant relative biases exhibited by the raw model predictions. Moreover, our results show that CNN-based downscaling yields better skill scores than the raw model forecasts over most portions of East Africa. The results demonstrate the potential usefulness of CNN in downscaling seasonal precipitation predictions over East Africa,particularly in providing improved forecast products which are essential for end users.

Growth and Interactions of Multi-Source Perturbations in Convection-Allowing Ensemble Forecasts

This study investigated the growth of forecast errors stemming from initial conditions(ICs),lateral boundary conditions(LBCs),and model(MO)perturbations,as well as their interactions,by conducting seven 36 h convectionallowing ensemble forecast(CAEF)experiments.Two cases,one with strong-forcing(SF)and the other with weak-forcing(WF),occurred over the Yangtze-Huai River basin(YHRB)in East China,were selected to examine the sources of uncertainties associated with perturbation growth under varying forcing backgrounds and the influence of these backgrounds on growth.The perturbations exhibited distinct characteristics in terms of temporal evolution,spatial propagation,and vertical distribution under different forcing backgrounds,indicating a dependence between perturbation growth and forcing background.A comparison of the perturbation growth in different precipitation areas revealed that IC and LBC perturbations were significantly influenced by the location of precipitation in the SF case,while MO perturbations were more responsive to convection triggering and dominated in the WF case.The vertical distribution of perturbations showed that the sources of uncertainties and the performance of perturbations varied between SF and WF cases,with LBC perturbations displaying notable case dependence.Furthermore,the interactions between perturbations were considered by exploring the added values of different source perturbations.For the SF case,the added values of IC,LBC,and MO perturbations were reflected in different forecast periods and different source uncertainties,suggesting that the combination of multi-source perturbations can yield positive interactions.In the WF case,MO perturbations provided a more accurate estimation of uncertainties downstream of the Dabie Mountain and need to be prioritized in the research on perturbation development.

Evaluating the Robustness of MDSS Maintenance Forecasts Using Connected Vehicle Data

The Indiana Department of Transportation (INDOT) adopted the Maintenance Decision Support System (MDSS) for user-defined plowing segments in the winter of 2008-2009. Since then, many new data sources, including connected vehicle data, enhanced weather data, and fleet telematics, have been integrated into INDOT winter operations activities. The objective of this study was to use these new data sources to conduct a systematic evaluation of the robustness of the MDSS forecasts. During the 2023-2024 winter season, 26 unique MDSS forecast data attributes were collected at 0, 1, 3, 6, 12 and 23-hour intervals from the observed storm time for 6 roadway segments during 13 individual storms. In total, over 888,000 MDSS data points were archived for this evaluation. This study developed novel visualizations to compare MDSS forecasts to multiple other independent data sources, including connected vehicle data, National Oceanic and Atmospheric Administration (NOAA) weather data, road friction data and snowplow telematics. Three Indiana storms, with varying characteristics and severity, were analyzed in detailed case studies. Those storms occurred on January 6th, 2024, January 13th, 2024 and February 16th, 2024. Incorporating these visualizations into winter weather after-action reports increases the robustness of post-storm performance analysis and allows road weather stakeholders to better understand the capabilities of MDSS. The results of this analysis will provide a framework for future MDSS evaluations and implementations as well as training tools for winter operation stakeholders in Indiana and beyond.

Short-Term Wind Speed Forecasts over the Pearl River Estuary:Numerical Model Evaluation and Deterministic Post-Processing

The Pearl River Estuary(PRE)is one of China’s busiest shipping hubs and fishery production centers,as well as a region with abundant island tourism and wind energy resources,which calls for accurate short-term wind forecasts.First,this study evaluated three operational numerical models,i.e.,ECMWF-EC,NCEP-GFS,and CMA-GD,for their ability to predict short-term wind speed over the PRE against in-situ observations during 2018-2021.Overall,ECMWF-EC out-performs other models with an average RMSE of 2.24 m s^(-1)and R of 0.57,but the NCEP-GFS performs better in the case of strong winds.Then,various bias correction and multi-model ensemble(MME)methods are used to perform the deterministic post-processing using a local and lead-specific scheme.Two-factor model output statistics(MOS2)is the optimal bias correction method for reducing(increasing)the overall RMSE(R)to 1.62(0.70)m s^(-1),demonstrating the benefits of considering both initial and lead-specific information.Intercomparison of MME results reveals that Multiple linear regression(MLR)presents superior skills,followed by random forest(RF),but it is slightly inferior to MOS2,particularly for the first few forecasting hours.Furthermore,the incorporation of additional features in MLR reduces the overall RMSE to 1.53 m s^(-1)and increases R to 0.74.Similarly,RF presents comparable results,and both outperform MOS2 in terms of correcting their deficiencies at the first few lead hours and limiting the error growth rate.Despite the satisfactory skill of deterministic post-processing techniques,they are unable to achieve a balanced performance between mean and extreme statistics.This highlights the necessity for further development of probabilistic forecasts.

Application Research of Temperature Forecasts on Elman Neural Network

[Objective] The aim was to establish Elman neural network model to predict the dynamic changes of temperature. [Method] Considering the inherent nature of temperature, and dy dint of the temperature in Chongqing during 1951-2010, the Elman artificial neural network model was applied to predict the temperature. [Result] This simulation result suggested that the relative error was small and can have a good simulation to the future temperature changes. [Conclusion] The prediction result can guide agricultural production and further apply to the field of pricing the weather derivative products.

A Model Output Machine Learning Method for Grid Temperature Forecasts in the Beijing Area

In this paper, the model output machine learning (MOML) method is proposed for simulating weather consultation, which can improve the forecast results of numerical weather prediction (NWP). During weather consultation, the forecasters obtain the final results by combining the observations with the NWP results and giving opinions based on their experience. It is obvious that using a suitable post-processing algorithm for simulating weather consultation is an interesting and important topic. MOML is a post-processing method based on machine learning, which matches NWP forecasts against observations through a regression function. By adopting different feature engineering of datasets and training periods, the observational and model data can be processed into the corresponding training set and test set. The MOML regression function uses an existing machine learning algorithm with the processed dataset to revise the output of NWP models combined with the observations, so as to improve the results of weather forecasts. To test the new approach for grid temperature forecasts, the 2-m surface air temperature in the Beijing area from the ECMWF model is used. MOML with different feature engineering is compared against the ECMWF model and modified model output statistics (MOS) method. MOML shows a better numerical performance than the ECMWF model and MOS, especially for winter. The results of MOML with a linear algorithm, running training period, and dataset using spatial interpolation ideas, are better than others when the forecast time is within a few days. The results of MOML with the Random Forest algorithm, year-round training period, and dataset containing surrounding gridpoint information, are better when the forecast time is longer.

Sensitivity to Tendency Perturbations of Tropical Cyclone Short-range Intensity Forecasts Generated by WRF

The present study uses the nonlinear singular vector(NFSV)approach to identify the optimally-growing tendency perturbations of the Weather Research and Forecasting(WRF)model for tropical cyclone(TC)intensity forecasts.For nine selected TC cases,the NFSV-tendency perturbations of the WRF model,including components of potential temperature and/or moisture,are calculated when TC intensities are forecasted with a 24-hour lead time,and their respective potential temperature components are demonstrated to have more impact on the TC intensity forecasts.The perturbations coherently show barotropic structure around the central location of the TCs at the 24-hour lead time,and their dominant energies concentrate in the middle layers of the atmosphere.Moreover,such structures do not depend on TC intensities and subsequent development of the TC.The NFSV-tendency perturbations may indicate that the model uncertainty that is represented by tendency perturbations but associated with the inner-core of TCs,makes larger contributions to the TC intensity forecast uncertainty.Further analysis shows that the TC intensity forecast skill could be greatly improved as preferentially superimposing an appropriate tendency perturbation associated with the sensitivity of NFSVs to correct the model,even if using a WRF with coarse resolution.

Improvement of 6–15 Day Precipitation Forecasts Using a Time-Lagged Ensemble Method

A time-lagged ensemble method is used to improve 6-15 day precipitation forecasts from the Beijing Climate Center Atmospheric General Circulation Model,version 2.0.1.The approach averages the deterministic predictions of precipitation from the most recent model run and from earlier runs,all at the same forecast valid time.This lagged average forecast （LAF） method assigns equal weight to each ensemble member and produces a forecast by taking the ensemble mean.Our analyses of the Equitable Threat Score,the Hanssen and Kuipers Score,and the frequency bias indicate that the LAF using five members at time-lagged intervals of 6 h improves 6-15 day forecasts of precipitation frequency above 1 mm d-1 and 5 mm d-1 in many regions of China,and is more effective than the LAF method with selection of the time-lagged interval of 12 or 24 h between ensemble members.In particular,significant improvements are seen over regions where the frequencies of rainfall days are higher than about 40％-50％ in the summer season; these regions include northeastern and central to southern China,and the southeastem Tibetan Plateau.

Effects of Dropsonde Data in Field Campaigns on Forecasts of Tropical Cyclones over the Western North Pacific in 2020 and the Role of CNOP Sensitivity

Valuable dropsonde data were obtained from multiple field campaigns targeting tropical cyclones,namely Higos,Nangka,Saudel,and Atsani,over the western North Pacific by the Hong Kong Observatory and Taiwan Central Weather Bureau in 2020.The conditional nonlinear optimal perturbation(CNOP)method has been utilized in real-time to identify the sensitive regions for targeting observations adhering to the procedure of real-time field campaigns for the first time.The observing system experiments were conducted to evaluate the effect of dropsonde data and CNOP sensitivity on TC forecasts in terms of track and intensity,using the Weather Research and Forecasting model.It is shown that the impact of assimilating all dropsonde data on both track and intensity forecasts is case-dependent.However,assimilation using only the dropsonde data inside the sensitive regions displays unanimously positive effects on both the track and intensity forecast,either of which obtains comparable benefits to or greatly reduces deterioration of the skill when assimilating all dropsonde data.Therefore,these results encourage us to further carry out targeting observations for the forecast of tropical cyclones according to CNOP sensitivity.

Sensitivities of Numerical Model Forecasts of Extreme Cyclone Events

A global forecast model is used to examine various sensitivities of numerical predictions of three extreme winter storms that occurred near the eastern continental margin of North America: the Ohio Valley blizzard of January 1978, the New England blizzard of February 1978, and the Mid-Atlantic cyclone of February 1979. While medium-resolution simulations capture much of the intensification, the forecasts of the precise timing and intensity levels suffer from various degrees of error. The coastal cyclones show a 5-10 hPa dependence on the western North Atlantic sea surface temperature, which is varied within a range (± 2.5℃) compatible with interannual fluctuations. The associated vertical velocities and precipitation rates show proportionately stronger dependences on the ocean temperature perturbations. The Ohio Valley blizzard, which intensified along a track 700-800 km from the coast, shows little sensitivity to ocean temperature. The effect of a shift of - 10?latitude in the position of the snow boundary is negligible in each case. The forecasts depend strongly on the model resolution, and the coarse-resolution forecasts are consistently inferior to the medium-resolution forecasts. Studies of the corresponding sensitivities of extreme cyclonic events over eastern Asia are encouraged in order to identify characteristics that are common to numerical forecasts for the two regions.

Bias correction of sea surface temperature retrospective forecasts in the South China Sea

Offline bias correction of numerical marine forecast products is an effective post-processing means to improve forecast accuracy. Two offline bias correction methods for sea surface temperature(SST) forecasts have been developed in this study: a backpropagation neural network(BPNN) algorithm, and a hybrid algorithm of empirical orthogonal function(EOF) analysis and BPNN(named EOF-BPNN). The performances of these two methods are validated using bias correction experiments implemented in the South China Sea(SCS), in which the target dataset is a six-year(2003–2008) daily mean time series of SST retrospective forecasts for one-day in advance, obtained from a regional ocean forecast and analysis system called the China Ocean Reanalysis(CORA),and the reference time series is the gridded satellite-based SST. The bias-correction results show that the two methods have similar good skills;however, the EOF-BPNN method is more than five times faster than the BPNN method. Before applying the bias correction, the basin-wide climatological error of the daily mean CORA SST retrospective forecasts in the SCS is up to-3°C;now, it is minimized substantially, falling within the error range(±0.5°C) of the satellite SST data.

Sensitivity of Medium-Range Weather Forecasts to the Use of Reference Atmosphere

In this paper, the authors develop the earlier work of Chen Jiabin et al. (1986). In order to reduce spectral truncation errors, the reference atmosphere has been introduced in ECMWF model, and the spectrally-represented variables, temperature, geopotential height and orography, are replaced by their deviations from the reference atmosphere. Two modified semi- implicit schemes have been proposed to alleviate the computational instability due to the introduction of reference atmosphere. Concerning the deviation of surface geopotential height from reference atmosphere, an exact computational formulation has been used instead of the approximate one in the earlier work. To re duce aliasing errors in the computations of the deviation of the surface geopotential height, a spectral fit has been used slightly to modify the original Gaussian grid-point values of orography.A series of experiments has been performed in order to assess the impact of the reference atmosphere on ECMWF medium- range forecasts at the resolution T21, T42 and T63. The results we have obtained reveal that the reference atmosphere introduced in ECMWF spectral model is generally beneficial to the mean statistical scores of 1000-200 hPa height 10-day forecasts over the globe. In the Southern Hemisphere, it is a clear improvement for T21, T42 and T63 throughout the 10-day forecast period. In the Northern Hemisphere, the impact of the reference atmos phere on anomaly correlation is positive for resolution T21, a very slightly damaging at T42 and almost neutral at T63 in the range of day 1 to day 4. Beyond the day 4 there is a clear improvement at all resolutions.

Customer Information Disclosure and Analyst Forecasts : Empirical Evidence from Analysts' Perspective

Based on the company's disclosure of key customer information,the impact of corporate customer concentration on analyst forecast was studied,and we further studied the impact of detailed customer names on analyst forecasts. It is found that:(i) customer concentration significantly affects the accuracy of analyst forecasts. The higher the customer concentration is,the lower the accuracy of analyst forecasts is;(ii) Voluntary disclosure of customer names can provide incremental information to analysts and mitigate the negative impact of customer concentration on the accuracy of analyst forests;(iii) further research has found that the incremental information brought by the state-owned enterprises' disclosure of the customer names to analysts is more obvious; disclosure of customer names by companies with high environmental uncertainty is more likely to be of concern to analysts; and star analysts have a higher ability to interpret customer names than non-star analysts.

Skill Assessment of Copernicus Climate Change Service Seasonal Ensemble Precipitation Forecasts over Iran

Medium to long-term precipitation forecasting plays a pivotal role in water resource management and development of warning systems.Recently,the Copernicus Climate Change Service(C3S)database has been releasing monthly forecasts for lead times of up to three months for public use.This study evaluated the ensemble forecasts of three C3S models over the period 1993-2017 in Iran’s eight classified precipitation clusters for one-to three-month lead times.Probabilistic and non-probabilistic criteria were used for evaluation.Furthermore,the skill of selected models was analyzed in dry and wet periods in different precipitation clusters.The results indicated that the models performed best in western precipitation clusters,while in the northern humid cluster the models had negative skill scores.All models were better at forecasting upper-tercile events in dry seasons and lower-tercile events in wet seasons.Moreover,with increasing lead time,the forecast skill of the models worsened.In terms of forecasting in dry and wet years,the forecasts of the models were generally close to observations,albeit they underestimated several severe dry periods and overestimated a few wet periods.Moreover,the multi-model forecasts generated via multivariate regression of the forecasts of the three models yielded better results compared with those of individual models.In general,the ECMWF and UKMO models were found to be appropriate for one-month-ahead precipitation forecasting in most clusters of Iran.For the clusters considered in Iran and for the long-range system versions considered,the Météo France model had lower skill than the other models.

A Spatiotemporal Interactive Processing Bias Correction Method for Operational Ocean Wave Forecasts

Numerical models and correct predictions are important for marine forecasting,but the forecasting results are often unable to satisfy the requirements of operational wave forecasting.Because bias between the predictions of numerical models and the actual sea state has been observed,predictions can only be released after correction by forecasters.This paper proposes a spati-otemporal interactive processing bias correction method to correct numerical prediction fields applied to the production and release of operational ocean wave forecasting products.The proposed method combines the advantages of numerical models and Forecast Discussion;specifically,it integrates subjective and objective information to achieve interactive spatiotemporal correc-tions for numerical prediction.The method corrects the single-time numerical prediction field in space by spatial interpolation and sub-zone numerical analyses using numerical model grid data in combination with real-time observations and the artificial judg-ment of forecasters to achieve numerical prediction accuracy.The difference between the original numerical prediction field and the spatial correction field is interpolated to an adjacent time series by successive correction analysis,thereby achieving highly efficient correction for multi-time forecasting fields.In this paper,the significant wave height forecasts from the European Centre for Medium-Range Weather Forecasts are used as background field for forecasting correction and analysis.Results indicate that the proposed method has good application potential for the bias correction of numerical predictions under different sea states.The method takes into account spatial correlations for the numerical prediction field and the time series development of the numerical model to correct numerical predictions efficiently.

Improvement in Background Error Covariances Using Ensemble Forecasts for Assimilation of High-Resolution Satellite Data

Satellite data obtained over synoptic data-sparse regions such as an ocean contribute toward improving the quality of the initial state of limited-area models. Background error covariances are crucial to the proper distribution of satellite-observed information in variational data assimilation. In the NMC （National Meteorological Center） method, background error covariances are underestimated over data-sparse regions such as an ocean because of small differences between different forecast times. Thus, it is necessary to reconstruct and tune the background error covariances so as to maximize the usefulness of the satellite data for the initial state of limited-area models, especially over an ocean where there is a lack of conventional data. In this study, we attempted to estimate background error covariances so as to provide adequate error statistics for data-sparse regions by using ensemble forecasts of optimal perturbations using bred vectors. The background error covariances estimated by the ensemble method reduced the overestimation of error amplitude obtained by the NMC method. By employing an appropriate horizontal length scale to exclude spurious correlations, the ensemble method produced better results than the NMC method in the assimilation of retrieved satellite data. Because the ensemble method distributes observed information over a limited local area, it would be more useful in the analysis of high-resolution satellite data. Accordingly, the performance of forecast models can be improved over the area where the satellite data are assimilated.

Artificial Neural Network for Combining Forecasts

This paper proposes artificial neural networks (ANN) as a tool for nonlinear combination of forecasts. In this study, three forecasting models are used for individual forecasts, and then two linear combining methods are used to compare with the ANN combining method. The comparative experiment using real--world data shows that the prediction by the ANN method outperforms those by linear combining methods. The paper suggests that the ANN combining method can be used as- an alternative to conventional linear combining methods to achieve greater forecasting accuracy.