Search

Short-term precipitation forecasting is essential for agriculture, transportation, urban management, and tourism. The radar echo extrapolation method is widely used in precipitation forecasting. To address issues like forecast degradation, insufficient capture of spatiotemporal dependencies, and low accuracy in radar echo extrapolation, we propose a new model: MS-DD3D-RSTN. This model employs spatiotemporal convolutional blocks (STCBs) as spatiotemporal feature extractors and uses the spatial-temporal loss (STLoss) function to learn intra-frame and inter-frame changes for end-to-end training, thereby capturing the spatiotemporal dependencies in radar echo signals. Experiments on the Sichuan dataset and the HKO-7 dataset show that the proposed model outperforms advanced models in terms of CSI and POD evaluation metrics. For 2 h forecasts with 20 dBZ and 30 dBZ reflectivity thresholds, the CSI metrics reached 0.538, 0.386, 0.485, and 0.198, respectively, representing the best levels among existing methods. The experiments demonstrate that the MS-DD3D-RSTN model enhances the ability to capture spatiotemporal dependencies, mitigates forecast degradation, and further improves radar echo prediction performance. [ABSTRACT FROM AUTHOR]

To evaluate the impact of different momentum control variable (CV) schemes (CV5, the momentum control variable option with ψχ and CV7, the momentum control variable option with UV) on radar data assimilation (DA) in weather research and forecasting model data-assimilation (WRFDA) systems, a heavy snowfall in central and eastern regions of China, which started on 6 February 2022, was taken as a case in this study. The results of the wind-field increments from the single observation tests indicated that the wind-field increments had a larger range of influence when stream function and velocity potential (ψχ) were used as momentum control variables in CV5. Some spurious increments were also generated in the wind-field analysis, since CV5 tended to maintain the integrated value of the wind field. When U-wind and V-wind were used as control variables in CV7, the wind-field increments had a smaller impact range, and there was less dependence among different locations on the wind increments. For the heavy snow case, the CV7 schemes displayed some improvements in simulating the composite reflectivity compared to the other two experiments, since the composite reflectivity in the CV5 and control experiments were overestimated to some level. It was also found that the RMSEs were lower in the CV7 compared to those in the CV5 in the short-term forecasts during the data-assimilation cycles. Results also indicated that the CV7 had a more significant effect on the 6 h accumulated precipitation forecasts. Meanwhile, the experiment Exp_CV7 achieved the best ETS and FSS scores among the three groups of experiments, while Exp_CV5 appeared to be generally superior to the CTRL. In summary, the precipitation of Exp_CV7 yielded the rainfall intensity and location most close to the observation compared to those from both the CTRL and Exp_CV5 experiments. [ABSTRACT FROM AUTHOR]

Precipitation forecasting is an immensely significant aspect of meteorological prediction. Accurate weather predictions facilitate services in sectors such as transportation, agriculture, and tourism. In recent years, deep learning-based radar echo extrapolation techniques have found effective applications in precipitation forecasting. However, the ability of existing methods to extract and characterize complex spatiotemporal features from radar echo images remains insufficient, resulting in suboptimal forecasting accuracy. This paper proposes a novel extrapolation algorithm based on a dual-branch encoder–decoder and spatiotemporal Gated Recurrent Unit. In this model, the dual-branch encoder–decoder structure independently encodes radar echo images in the temporal and spatial domains, thereby avoiding interference between spatiotemporal information. Additionally, we introduce a Multi-Scale Channel Attention Module (MSCAM) to learn global and local feature information from each encoder layer, thereby enhancing focus on radar image details. Furthermore, we propose a Spatiotemporal Attention Gated Recurrent Unit (STAGRU) that integrates attention mechanisms to handle temporal evolution and spatial relationships within radar data, enabling the extraction of spatiotemporal information from a broader receptive field. Experimental results demonstrate the model's ability to accurately predict morphological changes and motion trajectories of radar images on real radar datasets, exhibiting superior performance compared to existing models in terms of various evaluation metrics. This study effectively improves the accuracy of precipitation forecasting in radar echo images, provides technical support for the short-range forecasting of precipitation, and has good application prospects. [ABSTRACT FROM AUTHOR]

Spaceborne precipitation radars provide an unprecedented opportunity to study three-dimensional structure of precipitation, particularly over the open ocean where in-situ observations are rather meagre. In this study, instantaneous surface precipitation characteristics and their vertical structures during the tropical cyclones (TCs) over the North Indian Ocean (NIO) between 2014 and 2022 have been analysed using the dual-frequency precipitation radar onboard the Global Precipitation Measurement Core Observatory. Although stratiform precipitation accounts for more than 70% of total TC surface precipitation area, convective precipitation contributes about half of the total TC surface precipitation amount over the NIO. About 90% of stratiform TC precipitation area yields surface precipitation of less than 10 mm/hour. The vertical structures of stratiform and convective TC precipitation vary with surface precipitation intensity and have nearly similar characteristics over both basins of the NIO. This preliminary quantitative TC precipitation analysis would be useful for better understanding of precipitation processes during TCs over the NIO, and for further advancement in numerical models through improved parameterization schemes for TC precipitation forecasting. [ABSTRACT FROM AUTHOR]

During 6–7 May 2017, a record‐breaking nocturnal rainfall event occurred in Guangzhou, China, and it was a typical warm‐sector heavy rainfall event under weak synoptic forcing. A prior observational study by the authors revealed that warm‐rain microphysical processes were dominant and responsible for the record‐breaking precipitation. In this study, the double‐moment Morrison, Thompson and NSSL microphysics schemes in WRF are evaluated against polarimetric radar observations in their ability of reproducing observed microphysical characteristics. The Thompson scheme shows the greatest fidelity to the observed raindrop size distribution (RSD) median value, corresponding to the most amount of precipitation forecast. While the Morrison and NSSL simulations overestimate (underestimate) the raindrop size (number concentration), exhibiting continental‐type convective precipitation. The three experiments slightly overestimate differential reflectivity (ZDR), but significantly underestimate specific differential phase (KDP) and liquid water content by about 30%–50%, implying the undervaluation of number of medium‐sized raindrops. Examinations of the occurrence frequencies of ZDR, KDP, mass‐weighted diameter, and logarithmic normalized intercept parameter for rain suggest that all three schemes fail to reproduce the full variability of observed RSD for the extreme rainfall. The vertical variations of RSD parameters and the Kumjian‐Ryzhkov parameter space suggest that the collision–coalescence is the dominant warm‐rain microphysical process but the simulated process is too weak. This may be attributed to the misrepresented RSD near the melting layer, where the raindrops with lower number concentration and larger sizes cannot grow through the collision–coalescence process as actively. Plain Language Summary: Extreme rainfall is of great concern due to its potential to cause major disasters. Cloud microphysics has been considered one of the main sources of uncertainty in predicting rainfall at cloud‐resolving resolutions. This study investigates the performances of three commonly used schemes representing cloud precipitation processes in an extreme rainfall case that occurred in the warm and moist environment of South China. The Thompson scheme is found to perform better in simulating raindrop sizes and surface precipitation. None of the three schemes reproduces the full variation of observed raindrops, and all of them underestimate raindrop size growth through collision processes. These results suggest areas needing improvement in microphysics schemes for predicting extreme precipitation in sub‐tropical regions. Key Points: Thompson scheme better reproduces raindrop median size and number concentration than two other schemes in a Guangzhou extreme rain eventNone of the three schemes examined adequately predicts the full variability of raindrop size distribution important for extreme rainfallLow raindrop concentration and excessively large sizes in simulations jointly cause the too weak collision–coalescence processes [ABSTRACT FROM AUTHOR]

Dual-polarization radar data are useful for numerical models to improve precipitation forecasts. For an extremely heavy precipitation event that occurred in Central China on 11 August 2021, the hydrometeor concentration and water vapor content used in the initial field of the Weather Research and Forecasting (version 4.1) model are retrieved by the statistical relationship of relative humidity with dual-polarization radar reflectivity in Suizhou City of Central China. Three experiments are conducted, and the simulation results are compared after assimilating the radar data. The results indicate that the multiple factors contributing to this extreme heavy precipitation event included the divergence of upper-level airflows, the middle- and low-level low vortex/shear, the easterly jet stream in front of the low vortex, and the continuous intrusion of cold air on the ground. In addition, with the retrieval of the hydrometeor concentration and water vapor content, the composite reflectivity forecast results are more similar to the observations. Also, the location and intensity of the short-term extremely heavy precipitation event are less different from the observations. In addition, by cyclically adjusting the hydrometeor concentration and water vapor content in the initial field, we can obtain better forecasts of the reflectivity and short-term extremely heavy precipitation, and this improvement can be maintained for approximately 3 h. [ABSTRACT FROM AUTHOR]

This study assimilates for the first time polarimetric C-band radar observations from the German meteorological service (DWD) into DWD's convective-scale model ICON-D2 using DWD's ensemble-based KENDA assimilation framework. We compare the assimilation of conventional observations (CNV) with the additional assimilation of radar reflectivity Z (CNV + Z), with the additional assimilation of liquid or ice water content (LWC or IWC) estimates below or above the melting layer instead of Z where available (CNV + LWC /Z or CNV + IWC /Z respectively). Hourly quantitative precipitation forecasts (QPF) are evaluated for two stratiform and one convective rainfall events in the summers of 2017 and 2021. With optimized data assimilation settings (e.g., observation errors), the assimilation of LWC mostly improves first-guess QPF compared with the assimilation of Z alone (CNV + Z), whereas the assimilation of IWC does not, especially for convective cases, probably because of the lower quality of the IWC retrieval in these situations. Improvements are, however, notable for stratiform rainfall in 2021, for which the IWC estimator profits from better specific differential phase estimates owing to a higher radial radar resolution than the other cases. The assimilation of all radar data sets together (CNV + LWC + IWC + Z) yields the best first guesses. All assimilation configurations with radar information consistently improve deterministic 9 h QPF compared with the assimilation of only conventional data (CNV). Forecasts based on the assimilation of LWC and IWC retrievals on average slightly improve Fraction Skill Score (FSS) and Frequency Bias (FBI) compared with the assimilation of Z alone (CNV + Z), especially when LWC is assimilated for the 2017 convective case and when IWC is assimilated for the high-resolution 2021 stratiform case. However, IWC assimilation again degrades forecast FSS for the convective cases. Forecasts initiated using all radar data sets together (CNV + LWC + IWC + Z) yield the best FSS. The development of IWC retrievals that are more adequate for convection constitutes one next step to further improving the exploitation of ice microphysical retrievals for radar data assimilation. [ABSTRACT FROM AUTHOR]

Multiscale valid time shifting (VTS) was explored for a real-time convection-allowing ensemble (CAE) data assimilation (DA) system featuring hourly assimilation of conventional in situ and radar reflectivity observations, developed by the Multiscale Data Assimilation and Predictability Laboratory. VTS triples the base ensemble size using two subensembles containing member forecast output before and after the analysis time. Three configurations were tested with 108-member VTS-expanded ensembles: VTS for individual mesoscale conventional DA (ConVTS) or storm-scale radar DA (RadVTS) and VTS integrated to both DA components (BothVTS). Systematic verification demonstrated that BothVTS matched the DA spread and accuracy of the best-performing individual component VTS. The 10-member forecasts showed BothVTS performs similarly to ConVTS, with RadVTS having better skill in 1-h precipitation at forecast hours 1–6, while Both/ConVTS had better skill at later hours 7–15. An objective splitting of cases by 2-m temperature cold bias revealed RadVTS was more skillful than Both/ConVTS out to hour 10 for cold-biased cases, while BothVTS performed best at most hours for less-biased cases. A sensitivity experiment demonstrated improved performance of BothVTS when reducing the underlying model cold bias. Diagnostics revealed enhanced spurious convection of BothVTS for cold-biased cases was tied to larger analysis increments in temperature than moisture, resulting in erroneously high convective instability. This study is the first to examine the benefits of a multiscale VTS implementation, showing that BothVTS can be utilized to improve the overall performance of a multiscale CAE system. Further, these results underscore the need to limit biases within a DA and forecast system to best take advantage of VTS analysis benefits. [ABSTRACT FROM AUTHOR]

High-frequency cyclical assimilation of the retrieved rainwater and estimated in-cloud water vapor by radar reflectivity has positive impacts on convective precipitation forecasting but usually causes overestimation. The application of large-scale constraints will produce more balanced dynamical and thermal fields, which can address the above issue to some degree. In this study, the European Centre for Medium-Range Weather Forecasts (ECMWF) global forecast fields are utilized as large-scale constraints that are imposed on the regional model by the grid nudging method. Two heavy rainfall events that occurred in Jiangsu (the South case) and Hebei (the North case) Provinces with different water vapor background conditions are chosen. The results show that the experiment with dynamical constraints (nudging of the horizontal wind field only) performs the best 6-h precipitation location and intensity forecasts for both cases. The experiment that nudged the water vapor mixing ratio together with the horizontal wind field could significantly weaken the forecast precipitation intensity. Although it produces good precipitation forecasts in the first 3-h for the South case (under higher water vapor conditions), it produces an unreliable precipitation forecast with rapid decay for the North case. For the North case which is accompanied by significant cooling, the experiment nudging the water vapor mixing ratio, temperature and horizontal wind fields simultaneously performs better than the experiment nudging the water vapor mixing ratio together with the horizontal wind field. [ABSTRACT FROM AUTHOR]

This study presents a multiscale four-dimensional variational data assimilation (MS-4DVar) scheme that aims to assimilate multiscale information from conventional and radar observations. The MS-4DVar scheme separately assimilates conventional and radar data in different outer loop iterations of an incremental 4DVar with varied resolutions in the tangent linear and adjoint models (TLM/ADM) and time window lengths in the 4DVar. The MS-4DVar scheme was evaluated through a series of single observation tests and several cycled assimilation and forecasting experiments for a real squall-line case. Our results indicated that different TLM/ADM resolutions and time window lengths applied to the conventional and radar observations improved the multiscale analysis. In addition, the MS-4DVar scheme was more efficient than the common 4DVar because of the low-resolution TLM/ADM used for conventional data and the shortened time window length for radar data. Verification of the squall-line forecasts suggested that the MS-4DVar scheme improved the hourly accumulated precipitation and radar reflectivity forecast skills and reduced the forecast errors of both large-scale environmental and convective-scale states. Further diagnosis revealed that the improvement of precipitation forecast skill was attributable to the stronger cold pool, deeper saturated water vapor layer, and stronger updraft of the simulated squall-line system, as well as a more favorable convective environment. [ABSTRACT FROM AUTHOR]

A radar echo sequence which includes N frames plays a crucial role in monitoring precipitation clouds and serves as the foundation for accurate precipitation forecasting. To predict future frames, spatiotemporal models are used to leverage historical radar echo sequences. The spatiotemporal information combining both temporal and spatial information is derived from radar echo sequence. The spatiotemporal information reveals the changing trend of intensity in the echo region over time. Dynamic variation information extracted in radar echo maps mainly consists of nonstationary information and spatiotemporal information. However, the changing trends at different locations within the precipitation cloud are different, so the significance of the spatiotemporal information should be different. The current precipitation forecasting model, Memory In Memory (MIM), has the capability to store the nonstationary information derived from radar echo maps. However, the MIM falls short in discerning the significance of the spatiotemporal information extracted from these maps. To address this limitation, we propose a novel model, SAMM-MIM (self-attention memory module-MIM), which regulates the generation of hidden states and spatiotemporal memory states using a SAMM. The proposed SAMM uses the self-attention mechanism and a series of gate mechanisms to concentrate on significant spatiotemporal information, learn changing trends in the echo region, and output predictive information. The predictive information which is stored in hidden states contains predictions of the changing trends of dynamic variation information. Experimental evaluation on a dataset of radar data from Qingdao, China, demonstrates that SAMM-MIM achieves superior prediction performance compared with other spatiotemporal sequence models, as indicated by improved scores on mean squared error, critical success index, and missing alarm rate metrics. [ABSTRACT FROM AUTHOR]

Capabilities to directly assimilate radar data are implemented within the local ensemble transform Kalman filter (LETKF) and the gain‐form LETKF (LGETKF) algorithms of the Joint Effort for Data assimilation Integration (JEDI) system. The capabilities are evaluated for the analysis and forecast of a severe convection case of 20 May 2019 in the Southern Great Plains using the limited area model version of the FV3 dynamical core (FV3‐LAM) from a recent release for Short‐Range Weather Application (SRW App). The LETKF and LGETKF implementations are shown to produce analyses and short‐range forecasts comparable to those using the ensemble square‐root Kalman Filter (EnSRF) within the Gridpoint Statistical Interpolation (GSI) framework used by current NCEP operational models. In addition, LGETKF retaining only 60% variances for model‐space vertical localization performs similarly to LGETKF retaining 99% of variance and LETKF using observation error‐based vertical localization. JEDI LETKF shows better parallel scalability than LGETKF and GSI EnSRF. Plain Language Summary: The JEDI data assimilation (DA) framework under development will be used by all operational forecasting systems of the U.S. National Weather Service, including the Rapid Refresh Forecast System based on FV3, a dynamic core chosen for next‐generation NWS forecasting systems. In this study, we implemented radar DA capabilities into the JEDI with two local ensemble transform Kalman filter (LETKF) algorithms coupled with limited‐area FV3. Analyses and forecasts with radar DA better match radar reflectivity and radial velocity observations than forecasts without radar DA during the assimilation period. Three‐hour forecasts of precipitation, reflectivity, and updraft rotation show skills comparable to a DA experiment using the ensemble square‐root Kalman filter within the more mature GSI DA system. In addition, the new gain‐form LETKF, which has not been applied to radar DA before, performs quite well, although it is computationally more expensive than the regular LETKF. This work paves the way for using JEDI to assimilate radar data in operational forecasting systems. Key Points: Radar data assimilation capabilities are implemented and tested in Joint Effort for Data assimilation Integration (JEDI), a future data assimilation infrastructure for Unified Forecast System applicationsLocal ensemble transform Kalman filter (LETKF) algorithm with gain‐form vertical localization is tested for radar data assimilation and is found to perform similarly as LETKFThe enhanced gain‐form LETKF and LETKF in JEDI perform similarly as ensemble square‐root Kalman Filter within Gridpoint Statistical Interpolation for radar data assimilation [ABSTRACT FROM AUTHOR]

A sequence of radar echo maps can visually show the motion and variation trends of the echo area, making it a common tool for precipitation forecasting. The spatiotemporal context reveals the correlations of variation trends among different parts within the echo area. This paper proposes a novel precipitation forecasting model, ISTC-SA-MIM (Interactive Spatiotemporal Context Learning with Self-Attention and Memory in Memory), based on the MIM. Leveraging the spatiotemporal interactions and self-attention mechanism of the ISTC-SA structure, the proposed model effectively captures both long-term and short-term spatiotemporal contexts. By memorizing the spatiotemporal context and non-stationary information, ISTC-SA-MIM can accurately predict the motion and variation trends of the echo area. Radar echo data from the Qingdao station are collected as the dataset to evaluate the commonly used spatiotemporal models and ISTC-SA-MIM. The experiments demonstrate that ISTC-SA-MIM can predict the variation trends of the echo area more accurately by learning the spatiotemporal context. [ABSTRACT FROM AUTHOR]

This study explores the impact of assimilating radar radial velocity (RV) on the forecast of Super Typhoon Lekima (2019) using the Weather Research and Forecasting (WRF) model and three-dimensional variational (3DVAR) assimilation system with different background error length scales. The results of two single observation tests show that the smaller background error length scale is able to constrain the spread of radar observation information within a relatively reasonable range compared with the larger length scale. During the five data assimilation cycles, the position and structure of the near-land typhoon are found to be significantly affected by the setting of the background error length scale. With a reduced length scale, the WRF-3DVAR system could effectively assimilate the radar RV to produce more accurate analyses, resulting in an enhanced typhoon vortex with a dynamic and thermal balance. In the forecast fields, the experiment with a smaller length scale not only reduces the averaged track error for the 24-h forecasts to less than 20 km, but it also more accurately captures the evolutions of the typhoon vortex and rainband during typhoon landing. In addition, the spatial distribution and intensity of heavy precipitation are corrected. For the 24-h quantitative precipitation forecasts, the equitable threat scores of the experiment with a reduced length scale are greater than 0.4 for the threshold from 1 to 100 mm and not less than 0.2 until the threshold increases to 240 mm. The enhanced prediction performances are probably due to the improved TC analysis. [ABSTRACT FROM AUTHOR]

The performance of 4‐km Weather Research and Forecasting model in predicting reflectivity structure in China over three summer months is investigated using three‐dimensional (3D) reflectivity observations. Three verification domains, namely Southern China (SC), Central and Eastern China , and North China, that correspond to the three major rainfall centers of mainland China are selected. Results show that the forecasts reproduce the distribution and diurnal variation of precipitation well, but significant differences exist in the vertical distributions of the predicted and observed reflectivities. In observations, the highest frequency (of reflectivity ≥35 dBZ) occurs between 3 and 6 km, whereas it is at the surface in the forecast. The forecasts tend to over‐predict reflectivity intensity at the lower levels, especially in SC. Further evaluation using object‐based verification methods show that the forecasts greatly underestimate the afternoon peak frequency of precipitation clouds with reflectivity >30 dBZ. The forecasts fail to reproduce the diurnal variation of 35 dBZ mean and maximum height of the objects, producing less variation than observations. Analyses show that the failure in properly reproducing small‐scale reflectivity objects (with diameter <100 km) is primarily responsible for the underestimation of the mean and maximum object heights. Evaluation using additional 3D information show that the forecasts tend to produce a greater proportion of faster‐moving small‐scale objects. This study reveals that the simulation of the 3D structure of precipitation clouds in terms of reflectivity remains a great challenge, especially for smaller convective cells. Plain Language Summary: The simulation of precipitation clouds is associated carries the largest uncertainty in numerical weather prediction and climate models. With continued advancement in observations and numerical models, forecasts of horizontal distribution, intensity, diurnal variation and propagation of precipitation have been greatly improved. However, faithful representation of three‐dimensional (3D) cloud structure remains a great challenge, with the lack of systematic evaluations being one of the key causes. In this study, 3D gridded radar reflectivity observations are used to evaluate the precipitation clouds structure (with reflectivity ≥30 dBZ) predicted at a convection‐permitting‐resolution (CPR) over three summer months of 2016 in China. The evaluation reveals differences between the simulated and observed reflectivity structures, including the reflectivity intensity profile, the 35 dBZ mean and maximum height, the fraction of fast‐moving objects. CPR models still have difficulties in simulating 3D reflectivity structures especially for relatively small‐scale objects (with diameter <100 km). Key Points: Reflectivity structure predicted by the 4‐km Weather Research and Forecasting model are evaluated using 3D radar observationsThe model overpredicts the intensity of reflectivity at lower levels and fails to reproduce the diurnal variation of object heightThe failure to correctly reproduce the structure of smaller reflectivity objects is primarily responsible for the deviations [ABSTRACT FROM AUTHOR]

Data-driven machine learning technology can learn and extract features, a factor which is well recognized to be powerful in the warning and prediction of severe weather. With the large-scale deployment of the radar wind profile (RWP) observational network in China, dynamical variables with higher temporal and spatial resolution in the vertical become strong supports for machine-learning-based severe convection prediction. Based on the RWP mesonet that has been deployed in Beijing, this study uses the measurements from four triangles composed of six RWP stations to determine the profiles of divergence, vorticity, and vertical velocity before rainfall onsets. These dynamic feature variables, combined with cloud properties from Himawari-8 and ERA-5 reanalysis, serve as key input parameters for two rainfall forecast models based on the random forest (RF) classification algorithm. One is for the rainfall/non-rainfall forecast and another for the rainfall grade forecast. The roles of dynamic features such as divergence, vorticity, and vertical velocity are examined from ERA-5 reanalysis data and RWP measurements. The contribution of each feature variable to the performance of the RF model in independent tests is also discussed here. The results show that the usage of RWP observational data as the RF model input tends to result in better performance in rainfall/non-rainfall forecast 30 min in advance of rainfall onset than using the ERA-5 data as inputs. For the rainfall grade forecast, the divergence and vorticity that were estimated from the RWP measurements at 800 hPa show importance in improving the model performance in heavy and moderate rain forecasts. This indicates that the atmospheric dynamic variable measurements from RWP have great potential to improve the prediction skill of convection with the aid of a machine learning model. [ABSTRACT FROM AUTHOR]

Accurate and timely precipitation forecasts can help people and organizations make informed decisions, plan for potential weather-related disruptions, and protect lives and property. Instead of using physics-based numerical forecasts, which can be computationally prohibitive, there has been a growing interest in using deep learning techniques for precipitation prediction in recent years due to the success of these approaches in various other fields. These deep learning approaches generally use historical composite reflectivity (CR) at the surface level to predict future time steps. However, other relevant factors related to the potential motion and vertical structure of the storm have not been considered. To address this issue, this research proposes a multisource ConvLSTM (MS-ConvLSTM) model to improve the accuracy of precipitation forecasting by incorporating multiple data sources into the prediction process. The model was trained on a dataset of radar echo features, which includes not only composite reflectivity (CR), but also echo top (ET), vertically integrated liquid (VIL) water, and radar-retrieved wind field data at different elevations. Experiment results showed that the proposed model outperformed traditional methods in terms of various evaluation metrics, such as mean absolute error (MAE), mean squared error (MSE), probability of detection (POD), false alarm rate (FAR), and critical success index (CSI). [ABSTRACT FROM AUTHOR]

This study aims to develop and adapt new methodologies for the foundation of a new hydrological system, essential component of a much-needed decision support spatial system for the Firiza basin and reservoir (North-West of Romania) and other areas situated upstream of a densely populated area. In order to utilize semi-distributed models in basins outside of the USA, in particular the application of an adaptation of the methodology for application of gridded precipitation within HEC-HMS in US regions covered by the hydrologic rainfall analysis projection and the standard hydrologic grid systems has been developed for Romania. This methodology allows for precipitation processing and estimation of loss parameters and runoff transformation parameters to areas outside of the USA and areas that were not covered under the initial formulation of the method. This adaptation enables an alternative forecasting method and a controlled model for the transit of flash floods through reservoirs, which was correlated with the result of precipitation forecast. Multiple sets of parameters have been compiled to represent a wet/typical condition: a snowy and a dry one. All river basin conditions were adjusted according to historical events, and two zone configurations were developed, including the gauge configuration assigned to the transformation and basic flow parameters, and the soil configuration assigned to the loss parameters. Igniș radar provides radar imagery into the future, taking into account advection, growth and decay, allowing for an accurate 0–1 h quantitative precipitation forecast. This capability provides extremely accurate short-term forecasting—80% accuracy over 15 min and 70% accuracy over 30 min—that is vital for near-term flash flood forecasting. The efficiency of the model ranged between the good and very good. The results will be used to manage extreme situations such as flash floods, which can generate major hydrological risk situations for thousands of residents in the Baia Mare urban area, located downstream of this reservoir. [ABSTRACT FROM AUTHOR]

Radar reflectivity assimilation is often used to initialize hydrometeors, to which Numerical Weather Prediction (NWP) is highly sensitive. To better initialize hydrometeors, this study further developed the background error covariance (BEC) with vertical and multivariable correlations of hydrometeor control variables (H-BEC) in the WRF three-dimensional variational data assimilation system (WRFDA-3DVar). The impacts of the H-BEC are discussed using single radar reflectivity tests and series of cycling data assimilation and forecasting experiments for five multi-type convective rainfall cases. The conclusions are summarized as follows: (1) The vertical correlations can speed up the minimization of the cost function, whereas the multivariable correlations further accelerate this minimization; (2) The vertical correlations slightly improve the precipitation forecasting and only in the first hour, while multivariate correlations lead to a larger improvement and persist into the third hour; (3) The application of H-BEC leads to a more reasonable thermodynamic and dynamical structure of the initial field, thereby improving the capability of short-term precipitation forecasting. [ABSTRACT FROM AUTHOR]

Currently, most deep learning (DL)-based models for precipitation forecasting face two conspicuous issues: the smoothing effect in the precipitation field and the degenerate effect of forecasting precipitation intensity. Therefore, this study proposes "time series residual convolution (TSRC)", a DL-based convolutional neural network for precipitation nowcasting over China with a lead time of 3 h. The core idea of TSRC is it compensates the current local cues with previous local cues during convolution processes, so more contextual information and less uncertain features would remain in deep networks. We use four years' radar echo reflectivity data from 2017 to 2020 for model training and one year's data from 2021 for model testing and compare it with two commonly used nowcasting models: optical flow model (OF) and UNet. Results show that TSRC obtains better forecasting performances than OF and UNet with a relatively high probability of detection (POD), low false alarm rate (FAR), small mean absolute error (MAE) and high structural similarity index (SSIM), especially at longer lead times. Meanwhile, the results of two case studies suggest that TSRC still introduces smoothing effects and slightly outperforms UNet at longer lead times. The most considerable result is that our model can forecast high-intensity radar echoes even for typhoon rainfall systems, suggesting that the degenerate effect of forecasting precipitation intensity can be improved by our model. Future works will focus on the combination of multi-source data and the design of the model's architecture to gain further improvements in precipitation short-term forecasting. [ABSTRACT FROM AUTHOR]

Accurate short-term precipitation forecast is extremely important for urban flood warning and natural disaster prevention. In this paper, we present an innovative deep learning model named ISA-PredRNN (improved self-attention PredRNN) for precipitation nowcasting based on radar echoes on the basis of the advanced PredRNN-V2. We introduce the self-attention mechanism and the long-term memory state into the model and design a new set of gating mechanisms. To better capture different intensities of precipitation, the loss function with weights was designed. We further train the model using a combination of reverse scheduled sampling and scheduled sampling to learn the long-term dynamics from the radar echo sequences. Experimental results show that the new model (ISA-PredRNN) can effectively extract the spatiotemporal features of radar echo maps and obtain radar echo prediction results with a small gap from the ground truths. From the comparison with the other six models, the new ISA-PredRNN model has the most accurate prediction results with a critical success index (CSI) of 0.7001, 0.5812 and 0.3052 under the radar echo thresholds of 10 dBZ, 20 dBZ and 30 dBZ, respectively. [ABSTRACT FROM AUTHOR]

This work evaluates and compares the performance of the radar reflectivity and lightning data assimilation schemes implemented in weather research and forecasting-four-dimensional data assimilation (WRF-FDDA) for short-term precipitation and lightning forecasts. All six mesoscale convective systems (MCSs) with a duration greater than seven hours that occurred in the Guangdong Province of China during June 2020 were included in the experiments. The results show that both the radar reflectivity data assimilation and lightning data assimilation improved the analyses and short-term forecasts of the precipitation and lightning of the MCSs. On average, for precipitation forecasts, the experiments with radar reflectivity data assimilation performed better than those with lightning data assimilation; however, for lightning forecasts, the experiments with lightning data assimilation performed better in the analysis period and 1 h forecast, and for some cases, the superiority lasted to three forecast hours. This highlights the potential of lightning data assimilation in short-term lightning forecasting. [ABSTRACT FROM AUTHOR]

This study examined the impact of assimilating 3D temperature and water vapor information in addition to radar observations in a multiscale weather system. A frontal system with extremely heavy rainfall over northern Taiwan was selected. Using the WRF–LETKF Radar Assimilation System, we performed three sets of observing system simulation experiments to assimilate radar observations with or without thermodynamic variables obtained using different methods. First, assimilating the radar data for 2 h showed better structure and short-term forecast than 1 h. Second, we assimilated radar data and thermodynamic variables from a perfect model simulation. The results of the analysis revealed that when a precipitation position error occurred in the background field, assimilating thermodynamic information with the radar data could correct the dynamic structure and shorten the spinup assimilation period, resulting in substantial improvements to the quantitative precipitation forecast. Third, we applied a thermodynamics retrieval algorithm for a feasibility study. With a warm and wet bias of the retrieved fields, assimilating the temperature data had significant impact on the midlevel of stratiform areas and the forecast of the heavy rainfall was consequently improved. Assimilating the water vapor information helped reconstruct the range and intensity of the cold pool, but the improvement of rainfall forecast was limited. The optimal results of analysis and short-term forecast were achieved when both retrieved temperature and water vapor fields were assimilated. In conclusion, assimilating thermodynamic variables in the precipitation system is feasible for shortening the spinup period of data assimilation and improving the analysis and short-term forecast. [ABSTRACT FROM AUTHOR]

This work describes the implementation of an updated radar reflectivity assimilation scheme with the three-dimensional variational (3D-Var) system of Weather Research and Forecast (WRF). The updated scheme, instead of the original scheme assuming the relative humidity to a fixed value where radar reflectivity is higher than a threshold, assimilates pseudo water vapor retrieved by the Bayesian method, which would be consistent with clouds/precipitations provided by the model in theory. To verify the effect of the updated scheme to the improvement of precipitation simulation, a convective case in Wenquan County and the continuous monthly simulation with contrasting experiments in Xinjiang were performed. The test of single reflectivity observation demonstrates that the water vapor retrieved by the Bayesian method is consistent with the meteorological situation around. In the convective case, both the updated and original scheme results show that the assimilation of pseudo water vapor can adjust to the environmental conditions of water vapor and temperature. This can improve the hourly precipitation forecast skill more than the contrasting experiment, which was designed to only assimilate conventional observations and radar radial velocity data. In the continuous monthly experiments, the updated scheme reveals that the analysis of water vapor is more reasonable, and obtains a better precipitation forecast skill for 6 h accumulated precipitation than the contrasting experiments. [ABSTRACT FROM AUTHOR]

Radar data assimilation has been operational at the Deutscher Wetterdienst for several years and is essential for generating accurate precipitation forecasts. The current work attempts to further enhance the radar data assimilation by improving the latent heat nudging (LHN) scheme and by reducing the observation error (OE) caused by the representation error of the efficient modular volume radar operator (EMVORADO). First of all, a series of hindcasts for a one-month convective period over Germany are performed. Compared with radar reflectivity and satellite observations, it is found that the LHN scheme that implicitly adjusts temperature performs better, and the beam broadening effect and the choice of the scattering schemes in EMVORADO are important. Moreover, the Mie scheme with the new parameterization to reduce the brightband effect not only proves to be the best in hindcasts but also that it results in the smallest standard deviations and the shortest horizontal correlation length scales of the OE in data assimilation experiments. [ABSTRACT FROM AUTHOR]

Dual‐polarization radars can provide rich three‐dimensional (3D) information on cloud precipitation structure. To utilize the existing polarimetric radar network in operational data assimilation systems, a new polarimetric radar‐based cloud analysis method is introduced. New features include the employment of fuzzy‐logic hydrometeor classification, improved estimation of liquid and ice regions, and newly‐added number concentration estimation. The new scheme is evaluated with a typical squall line case. Results show that the 3D cloud precipitation structure and short‐term precipitation forecast is consistently improved. With extra information from the polarimetric observations, the new scheme is able to produce reasonable polarimetric signatures for analysis. More supercooled water results in more latent heat release which enhances the updraft, leading to stronger convection in the subsequent forecast. This study emphasizes the importance of correct initial liquid and ice particle condition for the prediction of deep convection. Plain Language Summary: Severe convective storms have always been one of the most the disastrous types of weather. However, due to their relatively small scale and rapid evolution, such storms are difficult to predict. Polarimetric radar networks have been established for nearly a decade and have been widely used in the monitoring and short‐term forecast of severe weather. Yet, this extra polarimetric information has not been used in the operational data assimilation system, which is key to a successful forecast. A new dual‐polarization (dual‐pol) radar‐based cloud analysis aiming to improve short‐term forecasts is therefore proposed. Results show that the new scheme is able to capture the 3D in‐cloud signature of deep convection. Consistent improvements are found for the prediction of 3D structure of deep convection and as well as for the quantitative precipitation forecast. Key Points: A modified cloud analysis model is proposed for polarimetric radar data assimilationA fuzzy‐logic method combined with ZDP is used to classify hydrometeor particles, followed by a partially double‐moment retrieval schemeExperiments with polarimetric‐data‐based cloud analysis show consistent improvement on analysis and forecast [ABSTRACT FROM AUTHOR]

A technique for model initialization using three-dimensional radar reflectivity data has been developed and applied within the NOAA 13-km Rapid Refresh (RAP) and 3-km High-Resolution Rapid Refresh (HRRR) regional forecast systems. This technique enabled the first assimilation of radar reflectivity data for operational NOAA forecast models, critical especially for more accurate short-range prediction of convective storms. For the RAP, the technique uses a diabatic digital filter initialization (DFI) procedure originally deployed to control initial inertial gravity wave noise. Within the forward-model integration portion of diabatic DFI, temperature tendencies obtained from the model cloud/precipitation processes are replaced by specified latent heating-based temperature tendencies derived from the three-dimensional radar reflectivity data, where available. To further refine initial conditions for the convection-allowing HRRR model, a similar procedure is used in the HRRR, but without DFI. Both of these procedures, together called the "Radar-LHI" (latent heating initialization) technique, have been essential for initialization of ongoing precipitation systems, especially convective systems, within all NOAA operational versions of the 13-km RAP and 3-km HRRR models extending through the latest implementation upgrade at NCEP in 2020. Application of the latent heat-derived temperature tendency induces a vertical circulation with low-level convergence and upper-level divergence in precipitation systems. Retrospective tests of the Radar-LHI technique show significant improvement in short-range (0-6 h) precipitation system forecasts, as revealed by reflectivity verification scores. Results presented document the impact on HRRR reflectivity forecasts of the radar reflectivity initialization technique applied to the RAP alone, HRRR alone, and both the RAP and HRRR. SIGNIFICANCE STATEMENT: The large forecast uncertainty of convective situations, even at short lead times, coupled with the hazardous weather they produce, makes convective storm prediction one of the most significant shortrange forecast challenges confronting the operational numerical weather prediction community. Prediction of heavy precipitation events also requires accurate initialization of precipitation systems. An innovative assimilation technique using radar reflectivity data to initialize NOAA operational weather prediction models is described. This technique, which uses latent heating specified from radar reflectivity (and can accommodate lightning data and other convection/precipitation indicators), was first implemented in 2009 at NOAA/NCEP and continues to be used in 2022 in the NCEP-operational RAP and HRRR models, making it a backbone of the NOAA rapidly updated numerical weather prediction capability. [ABSTRACT FROM AUTHOR]

This study quantitatively examines the contribution of assimilating observations in the regions with different dynamic instabilities to the analysis and prediction of an extreme rainstorm event in Fujian Province of China. The wind profiling radar (WPR) observations are classified into two groups, i.e., strong and weak instability areas (SIA and WIA), according to their local dynamic instability identified by the ensemble spread. Their performance of assimilation and prediction in terms of the wind and precipitation are evaluated and compared in detail. The results show that the wind analysis error by assimilating all of the WPR observations can be reduced by about 30%. In particular, the wind analysis errors by only assimilating the observations in the SIA are about 12% lower than those in the WIA. They are related to the existence of the low-level horizontal wind shear with strong instability in the SIA. The case study shows that the assimilation of observations in the SIA can effectively correct the wind fields on the two sides of the wind shear line, producing an improved precipitation forecast compared to observation assimilation in the WIA. [ABSTRACT FROM AUTHOR]

Based on the Weather Research and Forecasting (WRF) Model and the three-dimensional variational (3DVAR) data assimilation system, this study investigates the effects of assimilation of radar reflectivity and radial velocity under different momentum control variables on the forecast of Southwest China Vortex precipitation. It is shown that the U−V control variable strengthens the wind speed and vorticity to be better matching the observation, while using ψ−χ as the control variable will produce too large increments which are unphysical. The root mean square errors (RMSE) of radar radial velocity are around 2.4 m/s in the experiment using ψ−χ control variables, while the RMSE are below 2 m/s in the experiment with U−V control variables. The composite reflectivity from the analysis of the U−V control variables matches better with the observation than that from the analysis of the ψ−χ control variables, i.e., the forecast rain band location under U−V control variables is more accurate. ψ−χ control variable enhances the cold high-pressure system in near surface, while the U−V control variable is not significant. The water vapor flux convergence in the lower layers of the ψ−χ control variable is overestimated leading excessive precipitation in the forecast. The Equitable Threat Score (ETS) of the U−V control variable is about 0.1 higher than ψ−χ control variable. In summary, the U−V control variable is superior to the ψ−χ control variable in terms of analysis and forecasting about Southwest China Vortex precipitation. [ABSTRACT FROM AUTHOR]

An assumption of uncorrelated observation errors is commonly adopted in conventional data assimilation. For this reason, high‐resolution data are resampled with strategies such as superobbing or data thinning. These strategies diminish the advantages of high temporal and spatial resolutions that can provide essential details in convection development. However, assimilating high‐resolution data, such as radar radial wind, without considering observation error correlations can lead to overfitting and thus degrade the performance of data assimilation and forecasts. This study uses a radar ensemble data assimilation system that combines the Weather Research and Forecasting model and Local Ensemble Transform Kalman Filter (WRF‐LETKF) to assimilate radar radial wind and reflectivity data. We present a strategy to include the error correlation of the Doppler radar radial wind in the WRF‐LETKF radar assimilation system and examine its impact on the accuracy of short‐term precipitation predictions based on a heavy rainfall event on June 2, 2017 in Taiwan. For radial wind, the horizontal error correlation scale is approximately 25 km according to the innovation statistics. The introduction of observation error correlation for radar radial wind assimilation produces more small‐scale features in wind analysis corrections compared to the experiment using an independent observation assumption. Consequently, the modification of wind corrections leads to stronger convergence accompanied by higher water vapor content, which enhances local convections. This results in more accurate simulations of reflectivity and short‐term precipitation. In particular, this advantage is identified for extreme heavy rainfall thresholds at small scales according to probability quantitative precipitation forecasts and fractions skill score. [ABSTRACT FROM AUTHOR]

This study proposes a method for direct assimilation of radar reflectivity in cycled deterministic forecast systems using a time‐lagged ensemble based ensemble Kalman filter with the "cloud‐dependent" background error covariances. With this method, forecasts from multiple deterministic forecast cycles that are initialized at different times but valid at the same time, are collected as a time‐lagged ensemble to compute the background error covariances. An algorithm to compute the specified "cloud‐dependent" background error covariances was devised, in which the vertical background error covariance is computed for each of the "cloud‐feature" bins which are divided according to the column maximum radar reflectivity and cloud top height. The single‐observation experiments and the observing system simulation experiments (OSSEs) indicate that the radar reflectivity data assimilation method was able to capture the main thermodynamic and microphysical features of convective clouds and was effective in reducing the spin‐up time of convection, and generating convective cells which were missed in the background. Meanwhile, assimilating "zero‐reflectivity" observations helped suppress spurious convection. The method provided overall more accurate thermodynamic analysis, radar reflectivity and precipitation analysis and forecast of the MCS, while did not significantly increase computational cost, compared to a cloud‐analysis based latent‐heat nudging approach. Plain Language Summary: With a cycled forecast system, for a given time, several forecasts are produced by the model initialized at different times. Such forecasts can serve as a time‐lagged ensemble, in which the differences in initial conditions among different forecast cycles are used in lieu of uncertainty estimates of the initial atmospheric state. In this study, we propose a method for direct assimilation of radar reflectivity data in cycled deterministic forecast systems, using a time‐lagged ensemble based ensemble Kalman filter. Forecasts from multiple deterministic forecast cycles that are initialized at different times but valid at the same time are collected as a time‐lagged ensemble to compute the background error covariances. We devised an algorithm to compute the specified "cloud‐dependent" background error covariances, in which the vertical background error covariance is computed for each of the "cloud‐feature" bins which are divided according to the column maximum radar reflectivity and cloud top height. The assimilation experiments indicate that the radar reflectivity data assimilation scheme can improve the analysis and short‐term forecast of the mesoscale convective system without significantly increasing computational cost compared to a cloud‐analysis based latent nudging method. Key Points: This paper describe a scheme to directly assimilate radar reflectivity data for cycled deterministic forecast systemsTime‐lagged ensemble is used to compute the "cloud‐dependent" background error covariancesThis new radar reflectivity data assimilation scheme can improve the analysis and short‐term forecast of mesoscale convective system [ABSTRACT FROM AUTHOR]

Latent heating (LH) is an important quantity in both weather forecasting and climate analysis, being the essential factor driving convective systems. Yet, inferring LH rates from our current observing systems is challenging at best. For climate studies, LH has been retrieved from the Precipitation Radar on the Tropical Rainfall Measuring Mission (TRMM) using model simulations in the look-up table (LUT) that relates instantaneous radar profiles to corresponding heating profiles. These radars, first on TRMM and then Global Precipitation Measurement Mission (GPM), provide a continuous record of LH. However, temporal resolution is too coarse to have a significant impacts on forecast models. In operational forecast models such as High-Resolution Rapid Refresh, convection is initiated from LH derived from ground based radar. Despite the high spatial and temporal resolution of ground-based radars, one disadvantage of using these sources is that its data are only available over well observed land areas. This study develops a method to derive LH from the Geostationary Operational Environmental Satellite-16 (GOES-16) in near-real time. Even though the visible and infrared channels on the Advanced Baseline Imager (ABI) provide mostly cloud top information, rapid changes in cloud top visible and infrared properties, when formulated as a LUT similar to those used by the TRMM and GPM radars, can equally be used to derive LH profiles for convective regions based on model simulations with a convective classification scheme and channel 14 (11.2 µm) brightness temperatures. Convective regions detected by GOES-16 are assigned LH from the LUT, and they are compared with LH from the Next Generation Weather Radar (NEXRAD) and one of the Dual-frequency Precipitation Radar (DPR) products, the Goddard Convective-Stratiform Heating (CSH). LH obtained from GOES-16 show similar magnitude with NEXRAD and CSH, and vertical distribution of LH is also very similar with CSH. One month analysis of total LH from convective clouds from GOES-16 and NEXRAD shows good correlation between the two products. Finally LH profiles from GOES-16 and NEXRAD are applied to WRF simulations for convective initiation and their results are compared to investigate their impacts in precipitation forecasts. Results show that LH from GOES-16 have similar impacts as NEXRAD, and improves the forecast significantly. [ABSTRACT FROM AUTHOR]

This study investigated the effect of the assimilation of the S- and Ka-band dual‐wavelength-retrieved water vapor data with radial wind and reflectivity data. The vertical profile of humidity, which provides environmental information before precipitation occurs, was obtained at low levels and thinned into averaged and four-quadrant profiles. Additionally, the following two strategies were examined: 1) assimilation of water vapor data with radar data for the entire 2 h and 2) assimilation of water vapor data in the first hour, and radial velocity and reflectivity data in the second hour. By using the WRF local ensemble transform Kalman filter data assimilation system, three real cases of the Dynamics of the Madden–Julian Oscillation experiment were examined through a series of experiments. The analysis results revealed that assimilating additional water vapor data more markedly improved the analysis at the convective scale than assimilating radial wind and reflectivity data alone. In addition, the strategy of assimilating only retrieved water vapor data in the first hour and radial wind and reflectivity data in the second hour achieved the optimal analysis and subsequent very short-term forecast. The evaluation of quantitative precipitation forecasting demonstrated that assimilating additional retrieved water vapor data distinctly improved the rain forecast compared with assimilating radar data only. When moisture data were assimilated, improved nowcasting could be extended up to 4 h. Furthermore, assimilating moisture profiles into four quadrants achieved more accurate analysis and forecast. Overall, our study demonstrated that the humidify information in nonprecipitation areas is critical for further improving the analysis and forecast of convective weather systems. [ABSTRACT FROM AUTHOR]

Gangwon-do (GWD) has complex terrain and surface characteristics due to its location to the East Sea and the Taebaek Mountain range. This coastal location and rugged terrain can amplify snowfall mechanisms, making it challenging to accurately predict the amount and location. This study compares two methods for assimilating radar data and analyzed snowfall prediction results. The two methods compared are local ensemble transform Kalman filter (LETKF) and three-dimensional variational (3DVAR) data assimilation (DA). LETKF improved the water vapor amount and temperature using the covariance of the ensemble members, but 3DVAR improved the water vapor mixing ratio and temperature through an operator that assumed the atmosphere was saturated when reflectivity was above a certain threshold. In 2018, to understand the snowfall in GWD region and support the Pyeongchang Winter Olympic and Paralympic Games, a long-term heavy snow observation campaign was conducted. The International Collaborative Experiments for the 2018 Pyeongchang Olympic Games Projects (ICE-POP 2018) data are used to study and verify the numerical experiments. From the initial field verification using ICE-POP observation data (radiosonde), wind in LETKF was more accurately simulated than 3DVAR, but it underestimated the water vapor amount and temperature in the lower troposphere due to a lack of a water vapor and temperature observation operator. Snowfall in GWD was less simulated in LETKF, whereas snowfall of 10.0 mm or more was simulated in 3DVAR, resulting in an error of 2.62 mm lower than LETKF. The results signify that water vapor assimilation is important in radar DA and significantly impacts precipitation forecasts, regardless of the DA method used. Therefore, it is necessary to apply observation operators for water vapor and temperature in radar DA. [ABSTRACT FROM AUTHOR]

The aim of this study was to evaluate the impact of radar-detected radial velocity and reflectivity data assimilation (DA) on precipitation forecasts using the Weather Research and Forecasting (WRF) numerical model and its three-dimensional variational DA system (WRFDA-3DVAR). For this purpose, one heavy rain event in the city of Manaus in the Central Amazon on 25 December 2017 was analyzed. Numerical experiments were conducted using different initial conditions and both cyclical and non-cyclical assimilation processes. The evaluation of these experiments, conducted with and without radar DA, was based on comparisons between estimated radar precipitation and simulated precipitation. The results show that radial velocity plays a fundamental role in improving the representation of the convective system. Additionally, the best results were obtained when reflectivity was included alongside radial velocity in non-cyclical assimilation processes. Therefore, it can be concluded that the assimilation of radar data positively impacted the short-term forecast of precipitation associated with the studied convective system. [ABSTRACT FROM AUTHOR]

The local ensemble transform Kalman filter (LETKF) suggested by Hunt et al., 2007 is a very popular method for ensemble data assimilation. It is the operational method for convective‐scale data assimilation at Deutscher Wetterdienst (DWD). At DWD, based on the LETKF, three‐dimensional volume radar observations are assimilated operationally for the operational ICON‐D2. However, one major challenge for the LETKF is the situation where observations show precipitation (reflectivity) whereas all ensemble members do not show such reflectivity at a given point in space. In this case, there is no sensitivity of the LETKF with respect to the observations, and the analysis increment based on the observed reflectivity is zero. The goal of this work is to develop a targeted covariance inflation (TCI) for the assimilation of 3D‐volume radar data based on the LETKF, adding artificial sensitivity and making the LETKF react properly to the radar observations. The basic idea of the TCI is to employ an additive covariance inflation as entrance point for the LETKF. Here, we construct perturbations to the simulated observation which are used by the core LETKF assimilation step. The perturbations are constructed such that they exhibit a correlation between humidity and reflectivity. This leads to a change in humidity in such a way that precipitation is more likely to occur. We describe and demonstrate the theoretical basis of the method. We then present a case study where targeted covariance inflation leads to a clear improvement of the LETKF and precipitation forecast. All examples are based on the German radar network and the ICON‐D2 model over Central Europe. [ABSTRACT FROM AUTHOR]

The evolution of a heavy rainfall event that occurred on 19 August 2014 in northern Taiwan is investigated using observed data and analyzed using a newly developed system, namely, IBM_VDRAS. This system is based on a four-dimensional (4D) Variational Doppler Radar Assimilation System (VDRAS) that is capable of assimilating radar observations and surface station data over a complex terrain by adopting the Immersed Boundary Method (IBM). This event has precipitating processes and track different from those frequently observed in northern Taiwan. From the surface observations and the high spatiotemporal resolution analysis fields generated by IBM_VDRAS, it has been found that the rainfall process started with two single convective cells triggered by the interaction between land-sea breeze and terrain in two different cities (Taoyuan and Taipei). The outflow of one of the convective cells developed in Taoyuan City at an earlier time along with the outflow of another convective system that developed in the Taipei Basin; the former provided favorable conditions to intensify the latter. The enhanced major convective cell moved to the Taipei City metropolitan area and produced 80-mm precipitation within approximately 2.5 h. The kinematic, thermodynamic, and microphysical fields of the convective cells were analyzed in detail to explain the mechanisms that helped maintain the structure of the rainfall system. Sensitivity experiments of quantitative precipitation forecast revealed that the terrains prevent the location of major rainfall from shifting outside of the Taipei Basin. By assimilating the surface data, the model can better predict the rainfall position. [ABSTRACT FROM AUTHOR]

Based on a short-time heavy rainfall in Anhui and the weather research and forecasting (WRF) model, the water vapor in the initial field of the model is retrieved using the statistical relationships of the reflectivity factor from the Doppler weather radar with the relative humidity and hydrometeor. Three-dimensional variational (3DVAR) assimilation method is used to assimilate the radar reflectivity factor and radial velocity, and then the impact of assimilating retrieved water vapor on the analysis and forecast of the torrential rain is assessed. The results show that, after assimilating the retrieved water vapor, the water vapor field in the model is significantly improved. The water vapor content in the middle layer of the model in the analyzed field is increased, corresponding well with the convective region. Meanwhile, the precipitation distribution during this weather process is successfully simulated. The mesoscale characteristics are better presented by the imageries of radar reflectivity factor, and false echoes are partially reduced. Besides, the prediction of short-time heavy rainfall regions is closer to the actual observations. After assimilating the retrieved water vapor, the simulated one-hour accumulated rainfall is closer to the actual observation, and the fraction skill score (FSS) is higher. [ABSTRACT FROM AUTHOR]

Precipitation forecasting has always been a challenging problem. Currently, meteorological radar echo data are used widely in the field of precipitation forecasting. From the radar echo images, the current precipitation situation can be obtained. Compared with the actual image, however, the radar echo extrapolation image has the disadvantages of echo loss and inaccurate echo tracks, which leads to low accuracy of the precipitation forecast. A more effective network based on the New-RainNet module for precipitation forecasting is proposed. The network is constructed by stacking New-RainNet modules that are established by adding a convolutional block attention module and a flexible switch from Adam to SGD optimization algorithm to the original convolutional long short-term memory network. The short-term and impending precipitation forecast in the next 0 to 60 min was successfully realized on the radar echo dataset of the Hong Kong Observatory and achieves satisfactory results. Experiments show that the network surpasses other methods in both network convergence and prediction accuracy. In addition, the radar echo image generated by the network prediction is also better than other methods in visual quality as it retains the image details and is closer to the real radar echo image than other methods. [ABSTRACT FROM AUTHOR]

A new method has been developed to directly assimilate volume scans of radar reflectivity data with 4D‐Var in the Met Office Unified Model. The method has been demonstrated in the convective‐scale hourly‐cycling UKV forecast model. Reflectivity observations from 18 C‐band radars in the British Isles are assimilated. This article describes the method of observation processing and quality control, the observation operator, and assimilation method. The assimilation method uses a minimum threshold rainwater content in the forward operator to give sensitivity to reflectivity observations where there is no rain in the background. Furthermore, the use of the Huber norm in the observation penalty function allows the use of observations with large innovations in the assimilation. A change was made to the precipitation efficiency in the microphysics scheme of the linear perturbation forecast model to ensure stability of the scheme. A case‐study is presented which demonstrates how the inclusion of reflectivity observations enhances convergence through analysis increments to the wind field, leading to improvements to the location of convective precipitation features in the forecast. Two‐month trials for summer and winter seasons demonstrate significant improvements to rain forecasts in the nowcasting range. [ABSTRACT FROM AUTHOR]

In this study, we propose a volume‐to‐point framework for quantitative precipitation estimation (QPE) based on the Quantitative Precipitation Estimation and Segregation Using Multiple Sensor (QPESUMS) Mosaic Radar data set. With a data volume consisting of the time series of gridded radar reflectivities over the Taiwan area, we used machine learning algorithms to establish a statistical model for QPE in weather stations. The model extracts spatial and temporal features from the input data volume and then associates these features with the location‐specific precipitations. In contrast to QPE methods based on the Z–R relation, we leverage the machine learning algorithms to automatically detect the evolution and movement of weather systems and associate these patterns to a location with specific topographic attributes. Specifically, we evaluated this framework with the hourly precipitation data of 45 weather stations in Taipei during 2013–2016. In comparison to the operational QPE scheme used by the Central Weather Bureau, the volume‐to‐point framework performed comparably well in general cases and excelled in detecting heavy‐rainfall events. By using the current results as the reference benchmark, the proposed method can integrate the heterogeneous data sources and potentially improve the forecast in extreme precipitation scenarios. Plain Language Summary: Quantitative precipitation estimation (QPE) is a method of approximating the amount of rain that has fallen at a location or across a region. In most use cases, weather service providers use radar signals to estimate the amount of precipitation through the formula describing the relationship between radar reflectivity and the size of raindrop particles, Z–R relation. The state‐of‐the‐art QPE methods with adjusted Z–R relation are robust and accurate in general. Yet, they consider only the radar signal at a given location, which represented a point‐to‐point framework. In this study, we proposed a volume‐to‐point alternative that estimates the amount of rain by considering the signals in a broader spatial region and a longer time‐span. By using a deep neural network to process the large data volume, we demonstrated that the proposed method performed comparably well in general cases and excelled in detecting heavy‐rainfall events. This method could improve advanced warning for flash flooding and make water resource management more effective. In ongoing research, we seek to extend our approach to integrate the heterogeneous data sources and to be applied to precipitation forecasting. Key Points: We proposed a novel deep learning approach for estimating precipitation from a large data volume and demonstrated it with QPE from aggregated radar dataIn comparison to the operational QPE scheme, the proposed framework performed comparably well in general cases and excelled in detecting heavy‐rainfall eventsThe proposed framework can be further extended to integrate heterogeneous data sources and potentially improve the QPF in extreme precipitation scenarios [ABSTRACT FROM AUTHOR]

A radar reflectivity observation operator with ice-phase hydrometeors was employed for the assimilation of the real radar observations of a squall line on June 14, 2009, over Anhui and Jiangsu provinces in China. The threedimensional variational (3DVAR) data-assimilation system along with the Weather Research and Forecasting (WRF) model was applied. The impact of assimilating radar radial velocity and reflectivity is investigated by analysing the meteorological conditions for different stages of the squall line. The results show that the radar data-assimilation experiments adjust and improve the initial fields to some extent for the squall line with the cycling data-assimilation strategy. The analysed and predicted radar composite reflectivity is significantly improved with the use of the ice-phase hydrometeors observation operator. The improvements are illustrated in terms of the wind fields for both the surface and upper level, the radar reflectivity, the surface perturbation potential temperature and the surface perturbation pressure. In addition, the forecast skills of the precipitation are improved noticeably for both the location and the intensity after using the ice-phase hydrometeors radar observation operator. [ABSTRACT FROM AUTHOR]

The assimilation of radar reflectivity data is of great interest for numerical weather prediction (NWP) models at the convective scale since these observations are very dense both in space and time and they are related to microphysical prognostic variables. So far, the assimilation of reflectivity in operational frameworks has been limited to derived products, like precipitation rates and humidity profiles. This approach is also adopted at ARPAE-SIMC, the Hydro-Meteo-Climate Structure of the Regional Agency for Prevention, Environment and Energy of Emilia-Romagna region in Italy, where latent heat nudging (LHN) is employed to assimilate precipitation rates in the COSMO-2I model, the convection-permitting version of the regional model developed by the Consortium for Small-scale Modelling (COSMO) which is run operationally to provide high-resolution forecasts over Italy. However, to fully exploit the information contained in these observations, the whole reflectivity volume should be directly assimilated. Nevertheless, despite several promising studies, its implementation in an operational framework has not yet been achieved. In this study, a set-up designed to assimilate reflectivities operationally in the COSMO-2I model through a local ensemble transform Kalman filter (LETKF) has been evaluated over 37 days in 2018 and performed 303 forecasts. The comparison with the current operational set-up based on LHN reveals an average improvement in quantitative precipitation forecasts (QPFs) up to 7 hr, even if the impact on convective cases is much stronger than that observed in conditions of stratiform precipitation. Moreover, a small but positive impact is noticed for the RMSE of upper-air variables, while the impact on bias is mixed. Mixed results are also obtained when considering surface variables. In the light of the results of this study, a pre-operational parallel system in which COSMO-2I analyses are generated replacing LHN by the direct assimilation of reflectivity volumes was implemented in April 2020. [ABSTRACT FROM AUTHOR]

Accurate and timely precipitation forecasts are crucial for early warning. Rainfall nowcasting, the process of statistically extrapolating recent rainfall observations, is increasingly used for short‐term forecasting. Nowcasts are generally constructed with high‐resolution radar observations. As a proof of concept, we construct nowcasts with country‐wide rainfall maps estimated from signal level data of commercial microwave links (CMLs) for 12 summer days in the Netherlands. CML nowcasts compare well to radar rainfall nowcasts. Provided well‐calibrated CML rainfall estimates are employed, CML nowcasts can outperform unadjusted real‐time radar nowcasts for high rainfall rates, which are underestimated as compared to a reference. Care should be taken with the sensitivity of the advection field derivation to areas with low CML coverage and the inherent measurement scale of CML data, which can be larger than the application scale. We see potential for rainfall nowcasting with CML data, for example, in regions where weather radars are absent. Plain Language Summary: Capturing the amount, timing and location of rainfall are of key importance for water management and the agricultural sector. Nowcasting of rainfall, a computational approach where the most recent rainfall observations are spatially extrapolated, is increasingly used for short‐term rainfall forecasts. The required rainfall observations normally originate from weather radars, but it has previously been demonstrated that rainfall can also be estimated from rain‐induced signal attenuation from the commercial microwave links (CMLs) in cellular telecommunication networks, which are used worldwide. In this study, we use such estimates to construct rainfall nowcasts for 12 summer events in the Netherlands, which we compare to nowcasts using operational radars. The results demonstrate that there is potential for using rainfall estimates from CML data to nowcast rainfall up to a few hours ahead, for example, in (low‐ and middle‐income) regions where high‐resolution rainfall observations or weather radars are absent. Key Points: Probabilistic rainfall nowcasts are constructed with rainfall estimates from commercial microwave linksThe nowcasts are compared to radar rainfall nowcasts for 12 summer daysNowcast skill depends on link network density and is generally comparable to that of radar rainfall nowcasts [ABSTRACT FROM AUTHOR]

Weather radars and gauge observations are the primary observations to determine the coverage and magnitude of precipitation; however, radar and gauge networks have significant coverage gaps, which can underrepresent or even miss the occurrence of precipitation. This is especially noticeable in mountainous regions and in shallow precipitation regimes. The following study presents a methodology to improve spatial representations of precipitation by seamlessly blending multiple precipitation sources within the Multi-Radar Multi-Sensor (MRMS) system. A high spatiotemporal resolution multisensor merged quantitative precipitation estimation (QPE) product (MSQPE) is generated by using gauge-corrected radar QPE as a primary precipitation source with a combination of hourly gauge observations, monthly precipitation climatologies, numerical weather prediction short-term precipitation forecasts, and satellite observations to use in areas of insufficient radar coverage. The merging of the precipitation sources is dependent upon radar coverage based on an updated MRMS radar quality index, surface and atmospheric conditions, topography, gauge locations, and precipitation values. Evaluations of the MSQPE product over the western United States resulted in improved statistical measures over its individual input precipitation sources, particularly the locally gauge-corrected radar QPE. TheMSQPE scheme demonstrated its ability to sufficiently fill in areas where radar alone failed to detect precipitation due to significant beam blockage or poor coverage while minimizing the generation of false precipitation and underestimation biases that resulted from radar overshooting precipitation. [ABSTRACT FROM AUTHOR]

Skillful quantitative precipitation forecast using the numerical weather prediction model relies on an accurate estimate of the atmospheric state as an initial condition. Variational assimilation methods (VAR) have the potential to provide improved initial state estimation to the numerical weather prediction model using observations, prior data (background), and their respective error covariance. The quality of variational assimilation hinges on the background error statistics (BES) as it weights the error in prior state and determines the spread of assimilated observations in model space. Traditional approaches used to model stationary BES in a three‐dimensional variational assimilation system often fail to represent the model error in BES. In this study, we have proposed an ensemble method using Stochastically Perturbed Parameterization Tendency to represent the model error in BES. The characteristics of the proposed BES are compared with the traditional approaches using the National Meteorological Centre method for different control variables choices. We have further tested the performance of the proposed method in improving the skill of precipitation forecast for an extreme rainfall event, which caused devastating flood over Chennai city, India, on December 2015. Results demonstrate that the use of the proposed method results in better forecast skill of convective precipitation in terms of both position and intensity than traditional National Meteorological Centre‐based BES. Best results are obtained when zonal and meridional momentum control variables are used for modeling ensemble‐based BES. Key Points: New method is proposed to represent model error of background error statisticsPerformance of proposed method improves forecast skill for an extreme eventBest results are obtained when using zonal and meridional momentum control variables [ABSTRACT FROM AUTHOR]

This study examines the impacts of humidity adjustment in a cloud analysis system on the analysis and forecast of a squall line that occurred in southeast China on 23–24 April 2007. Radial velocity data are assimilated using the ARPS three‐dimensional variational system while reflectivity data are assimilated by a cloud analysis system. Experiments with two different humidity adjustment schemes are performed, with the original and enhanced versions. Another experiment does not adjust moisture. Both schemes generally decrease the humidity in front of the convective line and increase the humidity within the convective and stratiform regions of squall line compared to no humidity adjustment, and the original scheme produces the higher humidity within precipitation regions, especially the stratiform region. Both schemes improve the forecast of squall line structure, including the leading convective line, a transition zone, and a trailing stratiform region. Among the three experiments, the enhanced scheme produces the highest precipitation forecast skill. The latent heating rates are also diagnosed to investigate the microphysical responses to the humidity adjustment. The cooling outside of the observed precipitation regions corresponding to the humidity reduction also acts to suppress spurious precipitation. Water vapor condensation into cloud water and cloud water evaporation generally dominate the latent heating/cooling below the freezing level. Compared to the enhanced scheme, the original scheme releases much more latent heat in the middle troposphere, causing more warming. This is linked to the higher cloud water condensation rate, due to the higher amount of moisture addition/adjustment by the original scheme. Key Points: An enhanced vertical velocity‐dependent moisture adjustment scheme produces better squall line structures in both analysis and forecastThe "overwarming" issue associated with the original scheme is linked to more latent heat release associated with higher condensation rateCondensation into and evaporation of cloud water dominates latent heating/cooling below the freezing level [ABSTRACT FROM AUTHOR]

Severe weather nowcasting is a crucial mission of atmospheric science for the betterment of society to save life, limb, and property. In this study, composite radar data from the Central Weather Bureau of 16 typhoons are collected to examine the statistical performance of the McGill Algorithm for Precipitation nowcasting using Lagrangian Extrapolation (MAPLE) over Taiwan, an extrapolation algorithm that predicts future precipitation based on current radar echoes. In addition, instead of mixing the precipitation between radar extrapolation and numerical model forecast as in previous studies, a blending system is formed by synthesizing the wind information from model forecast with the echo extrapolation motion field via a variational algorithm to improve the nowcasting system. The statistical results of the radar echo extrapolation for 16 typhoon cases show that while the quantitative precipitation nowcasting skill can persist for up to 2 h, significant distortion for the rotational system is found after 2 h. On the other hand, the blending system helps to capture and maintain the rotation of typhoon rainband structures. The blending system extends the nowcasting skill by 1 h to a total of 3 h. Furthermore, the blending scheme performs especially well after the typhoon makes landfall in Taiwan. For disaster prevention and mitigation, this blending nowcasting technique may provide effective weather information immediately. [ABSTRACT FROM AUTHOR]

Short-term precipitation forecast in local areas based on radar reflectance images has become a hot spot issue in the meteorological field, which has an important impact on daily life. Recently, deep learning techniques have been applied to this field, and the effect is promoted remarkably compared with traditional methods. However, existing deep learning-based methods have not considered the problem that different areas and channels exert different influence on precipitation. In this paper, we propose to incorporate the multihead attention into a dual-channel neural network to highlight the key areas for precipitation forecast. Furthermore, to solve the problem of excessive loss of global information caused by the attention mechanism, the residual connection is introduced into the proposed model. Quantitative and qualitative results demonstrate that the proposed method achieves the state-of-the-art precipitation forecast accuracy on the radar echo dataset. [ABSTRACT FROM AUTHOR]

The aim of this study is to provide an evaluation of the impact of two largely used data assimilation techniques, namely three‐ and four‐dimensional variational data assimilation systems (3D‐Var and 4D‐Var), on the forecasting of heavy precipitation events using the Weather Research and Forecasting (WRF) model. For this purpose, two flash flood events in central Italy are analysed. The first occurred on September 14, 2012 during an Intensive Observation Period of the Hydrological cycle in the Mediterranean experiment (HyMeX) campaign, while the other occurred on May 3, 2018. Radial velocity and reflectivity acquired by C‐band weather radars at Mt. Midia (central Italy) and San Pietro Capofiume (northern Italy), as well as conventional observations (SYNOP and TEMP), are assimilated into the WRF model to simulate these damaging flash flood events. In order to evaluate the impact of the 3D‐Var and 4D‐Var assimilation systems on the estimation of short‐term quantitative precipitation forecasts, several experiments are carried out using conventional observations with and without radar data. Rainfall evaluation is performed by means of point‐by‐point and filtering methodologies. The results point to a positive impact of the 4D‐Var technique compared to results without assimilation and with 3D‐Var experiments. More specifically, the 4D‐Var system produces an increase of up to 22% in terms of the Fractions Skill Score compared to 3D‐Var for the first flash flood event, while an increase of about 5% is achieved for the second event. The use of a warm start initialization results in a considerable reduction in the spin‐up time and a significant improvement in the rainfall forecast, suggesting that the initial precipitation spin‐up problem still occurs when using 4D‐Var. [ABSTRACT FROM AUTHOR]