Hourly Forecast Research Articles

Hybrid real-time wave forecasting model combining Gaussian process regression and neural networks

To safely and effectively conduct offshore operations, a real-time wave forecast model that can accurately and rapidly forecast waves hours or days in advance is highly valuable. Recent studies have investigated models with neural networks (NNs) and have demonstrated their advantages in terms of forecasting speed and accuracy. However, NNs are limited by their requirement for a large training dataset, which can hinder the construction of the model if sufficient training data are unavailable. Therefore, this study constructed a hybrid real-time wave forecast model by combining an NN model and Gaussian process regression (GPR) model. The objective was to obtain reasonably accurate forecasts that were more accurate than those from individual NN models even when sufficient training data for the NN model was unavailable. To derive an optimal prediction, the values predicted by the NN and GPR models were combined by utilizing the standard deviation of the latter and considering the accuracy of each model. Consequently, predictions that were more accurate compared to those of the individual models for an entire year were obtained. This improvement was particularly significant for predicting relatively large wave heights, as indicated by the daily maximum wave heights. Additionally, various accuracy evaluation metrics demonstrated that despite using a small training dataset, the optimal prediction of the hybrid model was still superior to those of the individual models for predicting the daily maximum wave heights.

A bivariate simultaneous pollutant forecasting approach by Unified Spectro-Spatial Graph Neural Network (USSGNN) and its application in prediction of O3 and NO2 for New Delhi, India

Declining urban air quality affects socioeconomic stability, public health, and ecosystems and is demanding attention of the administration to address environmental sustainability goals. Given the effects of ozone, a greenhouse gas, on local climate and health, this study introduces a Unified Spectro-Spatial Graph Neural Network (USS-GNN) designed for simultaneous 24-hour forecasting of ozone and its precursor, nitrogen-dioxide, while addressing their chemical interactions and spatiotemporal dynamics. This model exploits the graph structure of atmospheric dynamics and mines high-level spatial, spectral, and physical features from atmospheric data through a Dot Product Edge Attention mechanism and a location-aware graph feature rewiring technique. The proposed model is developed for Indian capital city New Delhi, utilizes hourly observations for the years 2021 and 2022 and achieved R2 values of 0.650 and 0.618, RMSE of 13.950 and 16.120 μg/m3, MAE of 10.730 and 12.930 μg/m3 for ozone and nitrogen-dioxide respectively, outperforming state-of-the-art models. The model’s forecast analysis identified error-prone areas, effects of local meteorology, and pollutant interdependencies. An ablation study further detailed the impacts of graph operations on forecasts. Moreover, this study promotes the utility of bivariate modeling frameworks in improving urban pollution monitoring and supporting sustainable city management through data-driven policy implementations.

TPTNet: A Data‐Driven Temperature Prediction Model Based on Turbulent Potential Temperature

AbstractA data‐driven model for predicting the surface temperature using neural networks was proposed to alleviate the computational burden of numerical weather prediction (NWP). Our model, named TPTNet uses only 2 m temperature measured at the weather stations of the South Korean Peninsula as input to predict the local temperature at finite forecast hours. The turbulent fluctuation component of the temperature was extracted from the station measurements by separating the climatology component accounting for the yearly and daily variations. The effect of station altitude was then compensated by introducing a potential temperature. The resulting turbulent potential temperature (TPT) data at irregularly distributed stations were used as input for predicting the TPT at forecast hours through three trained networks based on convolutional neural network, Swin Transformer, and a graph neural network. By comparing the prediction performance of our network with that of persistence and NWP, we found that our model can make predictions comparable to NWP for up to 12 hr.

Recurrent neural network using LSTM for prediction of atmospheric pollutants in the State of Veracruz, Mexico

This article shows the use of a recurrent neural network LSTM whose objective was to make a 24-hour ozone forecast with data obtained from the INECC (National Institute of Ecology and Climate Change) and the validated database of the National Information System of Air Quality (SINAICA spanish acronym). The problem of air quality in the world is essential due to health problems. Currently, the techniques used to generate forecasts in monitoring stations are not accurate and do not allow generating alerts to avoid exposure to poor air quality. The state of Veracruz in Mexico has 7 air quality monitoring stations, showing that it is feasible to alternatively have a betterquality pollutant concentration forecasting system. The results obtained on the correlations of variables, although weak, positive or negative, allowed us to recognize their usefulness in the development of training in the RNN. The use of neural networks was demonstrated for the forecast of ozone concentration values in the metropolitan area of Poza Rica, using the LSTM using Python, Tensor Flow and Keras, with a good fit to the normalized pattern with an average RMSE of 0.0053 ppm of the days chosen for testing between the period from June 2019 to July 2020. Future work will consider PM10 and PM2.5.

Bias correction of wind power forecasts with SCADA data and continuous learning

Wind energy plays a critical role in the transition towards renewable energy sources. However, the uncertainty and variability of wind can impede its full potential and the necessary growth of wind power capacity. To mitigate these challenges, wind power forecasting methods are employed for applications in power management, electricity trading, or maintenance scheduling. In this work, we present, evaluate, and compare four machine learning-based wind power forecasting models. Our models correct and improve 48-hour forecasts extracted from a numerical weather prediction (NWP) model. The models are evaluated on datasets from a wind park comprising 65 wind turbines. The best improvement in forecasting error and mean bias was achieved by a convolutional neural network, reducing the average NRMSE down to 22%, coupled with a significant reduction in mean bias, compared to a NRMSE of 35% from the strongly biased baseline model using uncorrected NWP forecasts. Our findings further indicate that changes to neural network architectures play a minor role in affecting the forecasting performance, and that future research should rather investigate changes in the model pipeline. Moreover, we introduce a continuous learning strategy, which is shown to achieve the highest forecasting performance improvements when new data is made available.

Investigation of model forecast biases and skilful prediction for Assam heavy rainfall 2022

Extreme rainfall events (ERE) during the summer monsoon season have been occurring over most parts of India resulting in flooding and immense socio-economic loss. These extremes are becoming a frequent norm in the hilly and mountainous regions of the country such as Assam. Assam received one of the most historical EREs from 14–June 17, 2022. The present study analyses the performance of a suite of high-resolution ensemble model forecasts for this extreme event in terms of its intensity, and distribution with a lead time of up to 96 h. Furthermore, the 36 numerical experiments are carried out using two different land use and land cover (LULC) data sets (i.e. ISRO and USGS) and three different sets of parameterization schemes (i.e. planetary boundary layer, cumulus, and microphysics).Rainfall distributions in the case of USGS LULC are relatively less coherent and underestimated (60–260 mm/day) against IMD (80–300 mm/day) including the rainfall categories heavy (HR), very heavy (VHR), and extremely heavy (EHR) rainfall throughout the day-1 to day-4. Among all the ensembles (E1-E10), USGS (E6 - E10) has underestimated rainfall (140–260 mm/day) compared to ISRO (150–280 mm/day), specifically in MR and HR categories over the upper Assam (UAD) and lower Assam (LAD) divisions. Further, the Bias Correction Ensemble (BCE) technique is applied to minimize the forecast errors. A rigorous statistical analysis in terms of frequency distribution, Taylor diagram, and benchmark skill scores is carried out to elucidate the model biases. The set of the model ensembles using ISRO (E1- E5) and USGS (E6- E10) reasonably captured the HR, VHR, and EHR. In addition, throughout the forecast hour, BCE E5 (E10) is noted with the distinct realistic (underestimated) representation of model bias (5–20 %) (10–30 %) over all the subdivisions of Assam. Our results suggest that the combined efforts of ensembles of physical parameterization schemes, along with proper LULC, and the BCE approach are required to overcome challenges to improve the skills of rainfall events, particularly over complex terrains such as Assam.

Model validation and applications of wave and current forecasts from the Hong Kong Observatory's Operational Marine Forecasting System

The Hong Kong Observatory has been running an Operational Marine Forecasting System (OMFS) adapted from the Regional Ocean Modelling System (ROMS) coupled with the WaveWatch III and SWAN wave models to provide wave, current and sea temperature forecasts up to 144 h twice a day since December 2021. To facilitate users’ interpretation of model forecasts of significant wave height and current speed in coastal predictions and open seas which are of particular significance in high wind situations, model forecasts were validated against moored buoy observations and wave recorder measurements near the shores of Hong Kong and drifting buoy data over the South China Sea, as well as Mercator Ocean model reanalysis in 2022. The validation results showed that the wave forecasts generally agreed well with the buoy observations with coefficient of determination (R2) of around 0.7 and root-mean-square error (RMSE) of less than 0.2 m up to 72 h ahead. The R2 for sea current forecasts ranged between 0.4 and 0.6, and the RMSE was around 8 to 11 cm/s in near shores up to T + 144 forecast hours. Validation against drifting buoy demonstrated that the trend of current forecasts generally agreed well with the measurements. RMSE of surface current forecasts over open seas ranged from 19 cm/s for 24-hour forecast to around 30 cm/s for 144-hour forecast when compared against Mercator Ocean reanalysis. Results from the current downscaling approach could serve as a benchmark reference for HKO to enhance OMFS in the future. In this paper, applications of model forecasts in the provision of marine weather services in Hong Kong are also introduced.

Subgrid-Scale Topographic Effects on Radiation for Global Weather Forecast Models

The incoming solar radiation arriving the Earth’s surface is strongly influenced by surface terrain. Conventional global weather forecast models, with grid scales of about 10 km, lack the resolution to accurately capture terrain-induced variations. We devised a new parameterization method to incorporate high-resolution subgrid-scale terrain data into grid-scale radiative flux calculations without averaging or smoothing topographic features. Utilizing a 15″ digital elevation model from the Shuttle Radar Topography Mission, we computed subgrid-scale data and transformed them onto the Korean Integrated Model’s cubed sphere grid using a Voronoi diagram to maintain geographical accuracy. The new scheme was initially evaluated through offline ideal tests and case studies. The results demonstrated that the scheme accurately captured the variations in downward shortwave flux, keeping the mean flux on a global scale nearly constant. The global mean flux difference in all skies was less than 0.01%. Statistical analyses demonstrated improved temperature and geopotential height predictions compared to reanalysis data. The anomaly correlation coefficient for East Asia at 850 hPa increased by 0.036 at 240 forecast hours. Overall, the anomaly correlation coefficient and root mean square error of geopotential height and temperature showed enhancements, particularly in the Northern Hemisphere and tropics. Importantly, the scheme introduces negligible additional memory and CPU requirements, making it suitable for both regional and global models. Only a 0.58% increase in CPU time was observed for the 10-day forecast.

Assimilation of new rocket dropsonde data using WRFDA and its impact on numerical simulations of typhoon NORU

On September 26 at 2100 UTC and September 27 at 0900 and 2300 UTC, three rockets platform carrying dropsondes (TFTC-400) devices were launched off the east coast of Hainan Island to conduct a launch experiment aimed at detecting Typhoon NORU (2216). The experiment yielded valuable data that were subsequently analyzed to ascertain temperatures, wind speeds, and relative humidity in the atmosphere. Of the four experiments conducted, employing three distinct control variable configurations (CVs), we utilized the 3DVAR of WRF Data Assimilation (WRFDA) to assimilate rocket sounding data and the NCEP ADP Global Upper Air Observational Weather Data from the research data archive dataset that was jointly produced by the Center for Weather and Environmental Prediction (NCEP) and the National Center for Atmospheric Research (NCAR). In one experiment, no data assimilation was performed (CTL). These experiments were designed to assess the impact of these observational datasets on typhoon predictions of the Weather Research & Forecasting Model (WRF) numerical simulations. Utilizing the assimilated background field, a 24-hour forecast was conducted, and the assimilation simulation was analyzed with regard to typhoon path, intensity, precipitation, and improvements in the background field. The results reveal that, on average, the three-assimilation experiment led to a 30 % reduction in track error compared to the CTL. Additionally, the assimilation experiment for CV7 of control variable configurations brought the maximum wind speed closer to observed data than the CTL between 6 and 12 h. The TS (threat score) evaluation of simulated 24-hour precipitation in the model domain indicates that the three assimilation schemes exhibit a degree of improvement in the forecast scores for 24-hour cumulative typhoon precipitation. Nevertheless, the simulation results for minimum sea-level pressure are unsatisfactory. To establish statistical significance, additional cases within the relevant region are necessary for result validation.

Spatiotemporal Convolutional Approach for the Short-Term Forecast of Hourly Heavy Rainfall Probability Integrating Numerical Weather Predictions and Surface Observations

Abstract The accurate prediction of short-term rainfall, and in particular the forecast of hourly heavy rainfall (HHR) probability, remains challenging for numerical weather prediction (NWP) models. Here, we introduce a deep learning (DL) model, PredRNNv2-AWS, a convolutional recurrent neural network designed for deterministic short-term rainfall forecasting. This model integrates surface rainfall observations and atmospheric variables simulated by the Precision Weather Analysis and Forecasting System (PWAFS). Our DL model produces realistic hourly rainfall forecasts for the next 13 h. Quantitative evaluations show that the use of surface rainfall observations as one of the predictors achieves higher performance (threat score) with 263% and 186% relative improvements over NWP simulations for the first 3 h and the entire forecast hours, respectively, at a threshold of 5 mm h−1. Noting that the optical-flow method also performs well in the initial hours, its predictions quickly worsen in the final hours compared to other experiments. The machine learning model, LightGBM, is then integrated to classify HHR from the predicted hourly rainfall of PredRNNv2-AWS. The results show that PredRNNv2-AWS can better reflect actual HHR conditions compared with PredRNNv2 and PWAFS. A representative case demonstrates the superiority of PredRNNv2-AWS in predicting the evolution of the rainy system, which substantially improves the accuracy of the HHR prediction. A test case involving the extreme flood event in Zhengzhou exemplifies the generalizability of our proposed model. Our model offers a reliable framework to predict target variables that can be obtained from numerical simulations and observations, e.g., visibility, wind power, solar energy, and air pollution.

WoFS and the Wisdom of the Crowd: The Impact of the Warn-on-Forecast System on Hourly Forecasts during the 2021 NOAA Hazardous Weather Testbed Spring Forecasting Experiment

Abstract During the 2021 Spring Forecasting Experiment (SFE), the usefulness of the experimental Warn-on-Forecast System (WoFS) ensemble guidance was tested with the issuance of short-term probabilistic hazard forecasts. One group of participants used the WoFS guidance, while another group did not. Individual forecasts issued by two NWS participants in each group were evaluated alongside a consensus forecast from the remaining participants. Participant forecasts of tornadoes, hail, and wind at lead times of ∼2–3 h and valid at 2200–2300, 2300–0000, and 0000–0100 UTC were evaluated subjectively during the SFE by participants the day after issuance, and objectively after the SFE concluded. These forecasts exist between the watch and the warning time frame, where WoFS is anticipated to be particularly impactful. The hourly probabilistic forecasts were skillful according to objective metrics like the fractions skill score. While the tornado forecasts were more reliable than the other hazards, there was no clear indication of any one hazard scoring highest across all metrics. WoFS availability improved the hourly probabilistic forecasts as measured by the subjective ratings and several objective metrics, including increased POD and decreased FAR at high probability thresholds. Generally, expert forecasts performed better than consensus forecasts, though expert forecasts overforecasted. Finally, this work explored the appropriate construction of practically perfect fields used during subjective verification, which participants frequently found to be too small and precise. Using a Gaussian smoother with σ = 70 km is recommended to create hourly practically perfect fields in future experiments. Significance Statement This work explores the impact of cutting-edge numerical weather prediction ensemble guidance (the Warn-on-Forecast System) on severe thunderstorm hazard outlooks at watch-to-warning time scales, typically between 1 and 6 h of lead time. Real-time forecast products in this time frame are currently provided on an as-needed basis, and the transition to continuous probabilistic forecast products across scales requires targeted research. Results showed that hourly probabilistic participant forecasts were skillful subjectively and statistically, and that the experimental guidance improved the forecasts. These results are promising for the implementation and value of the Warn-on-Forecast System to provide improved hazard timing and location guidance within severe weather watches. Suggestions are made to aid future subjective evaluations of watch-to-warning-scale probabilistic forecasts.

The Application Research of FCN Algorithm in Different Severe Convection Short-Time Nowcasting Technology in China, Gansu Province

This study explores the application of the fully convolutional network (FCN) algorithm to the field of meteorology, specifically for the short-term nowcasting of severe convective weather events such as hail, convective wind gust (CG), thunderstorms, and short-term heavy rain (STHR) in Gansu. The training data come from the European Center for Medium-Range Weather Forecasts (ECMWF) and real-time ground observations. The performance of the proposed FCN model, based on 2017 to 2021 training datasets, demonstrated a high prediction accuracy, with an overall error rate of 16.6%. Furthermore, the model exhibited an error rate of 18.6% across both severe and non-severe weather conditions when tested against the 2022 dataset. Operational deployment in 2023 yielded an average critical success index (CSI) of 24.3%, a probability of detection (POD) of 62.6%, and a false alarm ratio (FAR) of 71.2% for these convective events. It is noteworthy that the predicting performance for STHR was particularly effective with the highest POD and CSI, as well as the lowest FAR. CG and hail predictions had comparable CSI and FAR scores, although the POD for CG surpassed that for hail. The FCN model’s optimal performances in terms of hail prediction occurred at the 4th, 8th, and 10th forecast hours, while for CG, the 6th hour was most accurate, and for STHR, the 2nd and 4th hours were most effective. These findings underscore the FCN model’s ideal suitability for short-term forecasting of severe convective weather, presenting extensive prospects for the automation of meteorological operations in the future.

Unravelling oceanic wave patterns: A comparative study of machine learning approaches for predicting significant wave height

Predicting significant wave height (SWH) is critical to ocean engineering, navigation, and renewable energy harvesting. This study comprehensively analyzes several machine learning (ML) and deep learning (DL) models. These include the adaptive neuro-fuzzy inference system with subtractive clustering ANFIS (SC), an artificial neural network with Bayesian regularization ANN (BR), bidirectional long-short-term memory (BiLSTM), bidirectional gated recurrent unit (BiGRU), and BiGRU with attention mechanism (AM), all employed for the prediction of SWH. The models were evaluated using data from three stations: a Canadian station in the North Atlantic Ocean and two Chinese stations in the East China Sea. Our study demonstrates the superior performance of the ANFIS (SC) model across all stations. For instance, the ANFIS model achieved an R2 value of 0.9979 at the Canadian station, outperforming the gated recurrent unit (GRU) based model with an R2 of 0.9127 from previous research. Similarly, at the Lianyungang (LYG) station, the ANFIS model achieved an R2 value of 0.9812, surpassing the performance of the GRU network used in prior studies, which reached an R2 of 0.6436 for 6-hour forecasts. Moreover, our study outperforms the convolutional neural network (CNN) based BiGRUAM model used for shale oil production prediction regarding R2 values. Furthermore, compared to the dynamic ensemble ESN model used for wave height prediction, our ANFIS model demonstrated more robust and consistent performance. Therefore, this research significantly contributes to the field of SWH prediction by presenting the superior performance of the ANFIS model across different geographical locations and conditions. This research opens new avenues for future research and practical applications in oceanography, weather forecasting, and renewable energy harvesting.

Evaluation of surface shortwave downward radiation forecasts by the numerical weather prediction model AROME

Abstract. With the worldwide development of the solar energy sector, the need for reliable surface shortwave downward radiation (SWD) forecasts has significantly increased in recent years. SWD forecasts of a few hours to a few days based on numerical weather prediction (NWP) models are essential to facilitate the incorporation of solar energy into the electric grid and ensure network stability. However, SWD errors in NWP models can be substantial. In order to characterize the performances of AROME in detail, the operational NWP model of the French weather service Météo-France, a full year of hourly AROME forecasts is compared to corresponding in situ SWD measurements from 168 high-quality pyranometers covering France. In addition, to classify cloud scenes at high temporal frequency and over the whole territory, cloud products derived from the Satellite Application Facility for Nowcasting and Very Short Range Forecasting (SAF NWC) from geostationary satellites are also used. The 2020 mean bias is positive, with a value of 18 W m−2, meaning that AROME on average overestimates the SWD. The root-mean-square error is 98 W m−2. The situations that contribute the most to the bias correspond to cloudy skies in the model and in the observations, situations that are very frequent (66 %) and characterized by an annual bias of 24 W m−2. Part of this positive bias probably comes from an underestimation of cloud fraction in AROME, although this is not fully addressed in this study due to the lack of consistent observations at kilometer resolution. The other situations have less impact on SWD errors. Missed cloudy situations and erroneously predicted clouds, which generally correspond to clouds with a low impact on the SWD, also have low occurrence (4 % and 11 %). Likewise, well-predicted clear-sky conditions are characterized by a low bias (3 W m−2). When limited to overcast situations in the model, the bias in cloudy skies is small (1 W m−2) but results from large compensating errors. Indeed, further investigation shows that high clouds are systematically associated with a SWD positive bias, while low clouds are associated with a negative bias. This detailed analysis shows that the errors result from a combination of incorrect cloud optical properties and cloud fraction errors, highlighting the need for a more detailed evaluation of cloud properties. This study also provides valuable insights into the potential improvement of AROME physical parametrizations.

Verification of the Global Forecast System, North American Mesoscale Forecast System, and High-Resolution Rapid Refresh Model Near-Surface Forecasts by Use of the New York State Mesonet

Abstract Global Forecast System (GFS), North American Mesoscale Forecast System (NAM), and High-Resolution Rapid Refresh (HRRR) 2-m temperature, 10-m wind speed, and precipitation accumulation forecasts initialized at 1200 UTC are verified against New York State Mesonet (NYSM) observations from 1 January 2018 through 31 December 2021. NYSM observations at 126 site locations are used to calculate standard error statistics (e.g., forecast error, root-mean-square error) for temperature and wind speed and contingency table statistics for precipitation across forecast hours, meteorological seasons, and regions. The majority of the focus is placed on the first 18 forecast hours to allow for comparison among all three models. A daily NYSM station-mean temperature error analysis identified a slight cold bias at temperatures below 25°C in the GFS, a cool-to-warm bias as forecast temperatures warm in the HRRR, and a warm bias at temperatures above 30°C in each model. Differences arise when considering temperature biases with respect to lead times and seasons. Wind speeds are overforecast at all ranges in each season, and forecast wind speeds ≥ 18 m s−1 are rarely observed. Performance diagrams indicate overall good forecast performance at precipitation thresholds of 0.1–1.5 mm, but with a high frequency bias in the GFS and NAM. This paper provides an overview of deterministic forecast performance across New York State, with the aim of sharing common biases associated with temperature, wind speed, and precipitation with operational forecasters and is the first step in developing a real-time model forecast uncertainty prediction tool.

Assimilation of Satellite Lightning Data in a Storm-Scale Numerical Weather Prediction System Using a 3D-EnVar Data Assimilation Scheme

Abstract The objective of this study is to prepare the assimilation of the Meteosat Third Generation Lightning Imager (MTG-LI) observations in the regional numerical weather prediction model AROME-France using a 3D-EnVar data assimilation algorithm. As opposed to the current operational configuration of AROME-France data assimilation system, the hydrometeor specific contents are added in the control variable and are thus updated during the assimilation process. Consequently, a lightning observation operator based on the specific contents of snow and graupel is used. As the first MTG satellite was launched in December 2022, the real MTG-LI observations are not available yet, and proxy spaceborne lightning observations generated from ground-based lightning detection network are used. We assess the forecast skills of experiments assimilating lightning in a close to operational configuration by comparing the results to a reference experiment that does not assimilate lightning. In the studied cases, the frequency bias and fraction skill score are improved for brightness temperatures forecasts lower than 280 K up to 4 h after the assimilation, implying a better description of the general cloud cover. The convective cores, identified by brightness temperatures lower than 220 K, are better captured in the first forecast hour but this improvement is not maintained through time and quickly fades.

A Comparison of 3DEnVar and 4DEnVar for Convective-Scale Direct Radar Reflectivity Data Assimilation in the Context of a Filter and a Smoother

Abstract The Gridpoint Statistical Interpolation (GSI)-based four- and three-dimensional ensemble–variational (4DEnVar and 3DEnVar) methods are compared as a smoother and a filter, respectively, for rapidly changing storms using the convective-scale direct radar reflectivity data assimilation (DA) framework. Two sets of experiments with varying DA window lengths (WLs; 20, 40, 100, and 160 min) and radar observation intervals (RIs; 20 and 5 min) are conducted for the 5–6 May 2019 case. The RI determines the temporal resolution of ensemble perturbations for the smoother and the DA interval for the filter spanning the WL. For experiments with a 20-min RI, evaluations suggest that the filter and the smoother have comparable performance with a 20-min WL; however, extending the WL results in the outperformance of the filter over the smoother. Diagnostics reveal that the degradation of the smoother is attributed to the increased degree of nonlinearity and the issue of time-independent localization as the WL extends. Evaluations for experiments with different RIs under the same WL indicate that the outperformance of the filter over the smoother diminishes for most forecast hours at thresholds of 30 dBZ and above when shortening the RI. Diagnostics show that more frequent interruptions of the model introduce model imbalance for the filter, and the increased temporal resolution of ensemble perturbations enhances the degree of nonlinearity for the smoother. The impact of model imbalance on the filter overwhelms the enhanced nonlinearity on the smoother as the RI reduces. Significance Statement The background uncertainties of rapidly changing storms suffer from fast error growth and high degrees of nonlinearities during the data assimilation (DA) period. Two variants of the ensemble-based DA method can account for such temporal evolution. The smoother uses background ensemble from multiple observation times over an assimilation period to estimate the propagation of statistics. The filter frequently calculates the statistics at multiple observation times over the same period. Current comparisons of the smoother and the filter were mostly performed using simple models; however, unknowns remain for convection-allowing forecasts with additional complexities. This study compares the filter and the smoother for the convective-scale analysis and prediction using a real-data study and finds that the comparison varies with the assimilation period and the observation interval.

Assessing RRFS versus HRRR in Predicting Widespread Convective Systems over the Eastern CONUS

Abstract This study provides a comparison of the operational HRRR version 4 and its eventual successor, the experimental Rapid Refresh Forecast System (RRFS) model (summer 2022 version), at predicting the evolution of convective storm characteristics during widespread convective events that occurred primarily over the eastern United States during summer 2022. In total 32 widespread convective events were selected using observations from the MRMS composite reflectivity, which includes an equal number of MCSs, quasi-linear convective systems (QLCSs), clusters, and cellular convection. Each storm system was assessed on four primary characteristics: total storm area, total storm count, storm area ratio (an indicator of mean storm size), and storm size distributions. It was found that the HRRR predictions of total storm area were comparable to MRMS, while the RRFS overpredicted total storm area by 40%–60% depending on forecast lead time. Both models tended to underpredict storm counts particularly during the storm initiation and growth period. This bias in storm counts originates early in the model runs (forecast hour 1) and propagates through the simulation in both models indicating that both miss storm initiation events and/or merge individual storm objects too quickly. Thus, both models end up with mean storm sizes that are much larger than observed (RRFS more so than HRRR). Additional analyses revealed that the storm area and individual storm biases were largest for the clusters and cellular convective modes. These results can serve as a benchmark for assessing future versions of RRFS and will aid model users in interpreting forecast guidance.

The impact of Indian radar and lightning data assimilation on the short‐range forecasts of heavy rainfall events

AbstractThe accurate prediction of high‐impact weather systems using cloud‐resolving models is still a challenge among researchers. This study evaluates the consistency of the combination of the three‐dimensional variational technique within the Gridpoint Statistical Interpolation assimilation system (GSI‐3DVAR) and nudging in the same modelling system on short‐range forecasts of three heavy rainfall events from the southwest monsoon season of 2021 over the Indian subcontinent. Three experiments have been conducted (i) control (CNTL): assimilated conventional and satellite observations; (ii) radar and lightning data assimilation (RLDA): assimilated radar reflectivity and lightning proxy reflectivity data along with all observations used in CNTL; and (iii) lightning data assimilation (LDA): same as RLDA but without the assimilation of radar data; particularly done to test the impact of assimilation of only lightning data. The model‐simulated rainfall is evaluated by using the Integrated Multi‐Satellite Retrievals (IMERG) for Global Precipitation Measurement (GPM IMERG) rainfall data. The intercomparison of LDA and RLDA for event 1 highlighted that both represent the convective regions reasonably better than CNTL, but RLDA outperforms LDA and thus further assimilation experiments are done with RLDA. RLDA provided reasonably accurate forecasts compared to CNTL, which is evident in the spatial distribution of rainfall and area‐averaged three‐hourly accumulated rainfall. Verification metrics for the three selected heavy rainfall events reveal that an optimal forecast performance (especially in the first six hours of free forecast) is obtained by the simulation with assimilation of radar and lightning data during the pre‐forecast period, through correcting the position and timing of convective centres. The probability of detection (POD) values are higher for light rainfall categories than for the heavy rain categories. POD values were higher in RLDA than CNTL throughout simulation for all three events. For all these three selected events, fractions skill scores (FSS) of RLDA are always better than CNTL with different neighbourhood sizes for different threshold values throughout the forecast period.

Construction of multiscale initial perturbations and the associated impacts on convection-permitting ensemble forecasting

To avoid selection of filtered scales in multiscale initial perturbations and to evaluate the roles of multiscale initial perturbations on the China Meteorological Administration-convection-permitting ensemble prediction system (CMA-CPEPS), we proposed a new scale-blending technique and then constructed a blended initial perturbation scheme (BLEND) based on the ensemble transform Kalman filter (ETKF) and dynamical downscaling (DOWN) schemes. First, the results revealed that the BLEND scheme can increase the small-scale (large-scale) perturbations of the DOWN (ETKF) scheme, with the multiscale characteristics. Second, for dynamical variables, the BLEND scheme can improve the under-dispersion of the ETKF scheme at all forecast hours, and also improve the over-dispersion of the DOWN scheme at the initial lead time. Additionally, the probabilistic forecasting skill of the BLEND scheme is similar to that of the DOWN and ETKF schemes. Third, for precipitation, the BLEND scheme can increase the ensemble spread and reduce the forecast error of the ETKF scheme at all forecast hours, and at 15–21 and 30–36 h, respectively. And the BLEND scheme can reduce the forecast error of the DOWN scheme for most forecast hours and neighborhood radii. Furthermore, the BLEND scheme can completely improve the Brier scores of the ETKF scheme, and improve those of the DOWN scheme for large neighborhood radii and the final 6–12 h. Therefore, it is desirable to construct multiscale initial perturbations in CPEPSs.