Search

Correcting the forecast bias of numerical weather prediction models is important for severe weather warnings. The refined grid forecast requires direct correction on gridded forecast products, as opposed to correcting forecast data only at individual weather stations. In this study, a deep learning method called CU-net is proposed to correct the gridded forecasts of four weather variables from the European Centre for Medium-Range Weather Forecast Integrated Forecasting System global model (ECMWF-IFS): 2-m temperature, 2-m relative humidity, 10-m wind speed, and 10-m wind direction, with a forecast lead time of 24 h to 240 h in North China. First, the forecast correction problem is transformed into an image-to-image translation problem in deep learning under the CU-net architecture, which is based on convolutional neural networks. Second, the ECMWF-IFS forecasts and ECMWF reanalysis data (ERA5) from 2005 to 2018 are used as training, validation, and testing datasets. The predictors and labels (ground truth) of the model are created using the ECMWF-IFS and ERA5, respectively. Finally, the correction performance of CU-net is compared with a conventional method, anomaly numerical correction with observations (ANO). Results show that forecasts from CU-net have lower root mean square error, bias, mean absolute error, and higher correlation coefficient than those from ANO for all forecast lead times from 24 h to 240 h. CU-net improves upon the ECMWF-IFS forecast for all four weather variables in terms of the above evaluation metrics, whereas ANO improves upon ECMWF-IFS performance only for 2-m temperature and relative humidity. For the correction of the 10-m wind direction forecast, which is often difficult to achieve, CU-net also improves the correction performance.

Using a high-density automatic weather stations (AWS) dataset of hourly rainfall observations, the present study investigates the relationship between rainfall and elevation in the Beijing area, and further proposes a rainfall amount dependent parameterized algorithm considering the elevation effect on rainfall on hourly timescale. The parameterization equation is defined as a segmented nonlinear model, which calculates the mountain rainfall as a function of valley rainfall amount. Results show that there exists an evident enhancement of rainfall amount by elevation effect in the Beijing area. In particular, larger rainfall amount is generally found in higher mountains, especially for slight rain and moderate rain. Furthermore, six representative station pairs located in valleys and on mountains respectively are selected to estimate the values of optimal parameters in the parameterization equation. The parameterization algorithm of elevation dependence can produce a reduction in the root-mean-square error and obtain a much closer mountain rainfall total to the observations compared with those using no elevation dependence. Furthermore, the spatial distribution of rainfall is more realistic and accurate in mountainous terrain when elevation dependence is considered. This study helps to understand the variability of rainfall with complex terrain in the Beijing area, and gives a possible way to parameterize rainfall-elevation relationship on hourly timescale.

The seasonal variations of raindrop size distribution (DSD) and rainfall are investigated using three-year (2016–2018) observations from a two-dimensional video disdrometer (2DVD) located at a suburban station (40.13° N, 116.62° E, ~30 m AMSL) in Beijing, China. The annual distribution of rainfall presents a unimodal distribution with a peak in summer with total rainfall of 966.6 mm, followed by fall. Rain rate (R), mass-weighted mean diameter (Dm), and raindrop concentration (Nt) are stratified into six regimes to study their seasonal variation and relative rainfall contribution to the total seasonal rainfall. Heavy drizzle/light rain (R2: 0.2~2.5 mm h−1) has the maximum occurrence frequency throughout the year, while the total rainfall in summer is primarily from heavy rain (R4: 10~50 mm h−1). The rainfall for all seasons is contributed primarily from small raindrops (Dm2: 1.0~2.0 mm). The distribution of occurrence frequency of Nt and the relative rainfall contribution exhibit similar behavior during four seasons with Nt of 10~1000 m−3 registering the maximum occurrence and rainfall contributions. Rainfall in Beijing is dominated by stratiform rain (SR) throughout the year. There is no convective rainfall (CR) in winter, i.e., it occurs most often during summer. DSD of SR has minor seasonal differences, but varies significantly in CR. The mean values of log10Nw (Nw: mm−1m−3, the generalized intercept parameter) and Dm of CR indicate that the CR during spring and fall in Beijing is neither continental nor maritime, at the same time, the CR in summer is close to the maritime-like cluster. The radar reflectivity (Z) and rain rate (?) relationship (Z = ?R?) showed seasonal differences, but were close to the standard NEXRAD Z-R relationship in summer. The shape of raindrops observed from 2DVD was more spherical than the shape obtained from previous experiments, and the effect of different axis ratio relations on polarimetric radar measurements was investigated through T-matrix-based scattering simulations.

The Beijing Broadband Lightning Network (BLNET) was successfully set up in North China and had yielded a considerable detection capability of total lightning (intracloud and cloud to ground) over the regions with complex underlying (plains, mountains, and oceans). This study set up a basic framework for the operational application of assimilating total lightning activities from BLNET and assesses the potential benefits in cloud-scale, very short-term forecast (nowcasting) by modulating the vertical velocity using the 4DVar technique. Nowcast statistics aggregated over 11 cycles show that the nowcasting performances with the assimilation of BLNET lightning datasets outperform RAD and the assimilation of GLD360 (Global Lightning) datasets. The assimilation of BLNET data improves the model's dynamical states in the analysis by enhancing the convergence and updraft in and near the convective system. To better implement of assimilating real-time lightning data, this study also conducts sensitivity experiments to investigate the impact of the horizontal length scale of a distance-weighted interpolation, binning time intervals, and different vertical profile or distance weights prior to the DA. The results indicate that the best forecast performance for assimilating BLNET lightning datasets is obtained in a 4DVar cycle when the lightning accumulation interval is 3 min, the radius of horizontal interpolation is 5 × 5, and the statistically vertical velocity profile and the distance weights obtained from cumulus cloud.

A case study is presented of convection initiation (CI) resulting from the merger of a cold front with a dryline in southwestern Beijing, China, on the afternoon of 11 June 2011. This process is analyzed with S-band Doppler radar data, surface automatic weather station data, and mesoscale numerical simulation results. The formation of this dryline is analogous to that on the Great Plains of the United States, and it is conducive to CI with mesoscale updrafts generated from the baroclinic frontogenesis, and with favorable instability immediately on the moist side. Prior to the front–dryline merger, as the cold front approached the observed boundary layer convergence line, or the simulated meso- γ-scale secondary dryline, CI occurred ahead of the cold front with little contribution from frontogenetic baroclinity of the dryline. The cold front then merged with the dryline, and the baroclinity of the dryline was enhanced by the associated convergence, to a degree comparable to that caused by frontogenesis of the dryline itself, thus leading to more CI. During the front–dryline merger, meso- γ-scale discrete cold pools associated with the cold front led to a diverse distribution of CI.

The metropolis of Beijing in China is located on a plain adjacent to high mountains to its northwest and the gulf of the Bohai Sea to its southeast. One of the most challenging forecast problems for Beijing is to predict whether thunderstorms initiating over the mountains will propagate to the adjacent plains and intensify. In this study, 18 warm season convective cases between 2008 and 2013 initiating on the mountains and intensifing on the plains under weak synoptic forcing were analyzed to gain an understanding of their characteristics. The statistical analysis was based on mosaic reflectivity data from six operational Doppler radars and reanalysis data produced by the Four-Dimensional Variational Doppler Radar Analysis System (VDRAS). The analysis of the radar reflectivity data shows that convective precipitation strengthened on the plains at certain preferred locations. To investigate the environmental conditions favoring the strengthening of the mountain-to-plain convective systems, statistical diagnoses of the rapid-update (12 min) 3 km reanalyses from VDRAS for the 18 cases were performed by computing the horizontal and temporal means of CAPE (Convective Available Potential Energy), CIN (Convective Inhibition), vertical wind shear, and low-level wind for the plain and mountain regions separately. The results were compared with those from a baseline representing the warm season average and from a set of null cases and found considerable differences in these fields between the three datasets. The mean distributions of VDRAS reanalysis fields were also examined. The results suggest that the convergence between the low-level outflows associated with cold pools and the south-southeasterly environmental flows correspond well with the preferred locations of convective intensification on the plains.

With the rapidly growing number of connected smart devices deployed and diverse services provided in the Internet of Things (IoT), selecting proper services for users is becoming more and more important. However, challenges exist as a result of the highly heterogeneous environments, characteristics of various kinds of users and the myriad services offered by many service providers, which have promising applications in the IoT era. In the meantime, users' contexts (e.g., location, time, and surroundings) are wildly utilized in the IoT scenario to better satisfy individuals' demands, raising privacy issues among people. To address these problems, we proposed a differentially private tree-based contextual online learning approach for IoT service selection to select suitable services for users. Leveraging on the historical records of services' and users' feedback, our algorithm achieves high prediction accuracy. Besides, instead of considering the services as individual items, we utilize a top-down cover tree structure to select services, which supports increasing large-scale dataset and diverse natural conditions. We theoretically prove that the accumulative regret of our approach has a sublinear bound and our experiment confirms that it can handle big data problems while achieving a balance between privacy-preserving level and service selection accuracy.

The local convection initiation (CI) mechanisms of a convective case that occurred on 5 August 2017 in Cangzhou, northern China, were studied using Doppler radar and automatic weather station observational analysis, along with Variational Doppler Radar Analysis System assimilation analysis. During the convective process, a gust front appeared ahead of two existing convective systems, respectively. In the warm and moist environment ahead of the gust fronts in the south, there was a mesoscale air mass boundary. With the process of a gust front moving southward, approaching the mesoscale air mass boundary, the convergence intensified in the area between the gust front and the mesoscale air mass boundary. Finally, the strong convergent updraft exceeded the level of free convection and triggered the new convection.

Mechanical faults of high-voltage circuit breakers (HVCBs) always happen over long-term operation, so extracting the fault features and identifying the fault type have become a key issue for ensuring the security and reliability of power supply. Based on wavelet packet decomposition technology and random forest algorithm, an effective identification system was developed in this paper. First, compared with the incomplete description of Shannon entropy, the wavelet packet time-frequency energy rate (WTFER) was adopted as the input vector for the classifier model in the feature selection procedure. Then, a random forest classifier was used to diagnose the HVCB fault, assess the importance of the feature variable and optimize the feature space. Finally, the approach was verified based on actual HVCB vibration signals by considering six typical fault classes. The comparative experiment results show that the classification accuracy of the proposed method with the origin feature space reached 93.33% and reached up to 95.56% with optimized input feature vector of classifier. This indicates that feature optimization procedure is successful, and the proposed diagnosis algorithm has higher efficiency and robustness than traditional methods.

This paper analyses the transient characteristics and breaking performance of direct-current (DC) forced-interruption vacuum interrupters in 270 V power-supply systems. Three stages are identified in forced interruption: the DC-arcing stage, current-commutation stage, and voltage-recovery stage. During the current-commutation stage, the reverse peak-current coefficient k, which is a key design factor, is used to calculate the rate of current at zero-crossing (di/dt). MATLAB/Simulink simulation models are established to obtain the transient characteristics influenced by the forced-commutation branch parameters and the coefficient k. To study the breaking performance of spiral-type contacts, experiments are conducted for different contact materials and arcing times for currents less than 3.5 kA. During the DC-arcing stage, a locally intensive burning arc is observed in the CuW80 contact; however, it is not observed in the CuCr50 contact. On examining the re-ignition interruption results of the CuW80 contact, the intensive burning arc is found to be positioned within a possible re-ignition region. When the arcing time is longer than 1 ms, the intensive burning arc occurs and affects the breaking performance of the spiral-type contacts. If the DC-arcing stage is prolonged, the total arcing energy increases, which leads to a lower breaking capacity.

In this paper, we present our research into the interruption performance of vacuum circuit breakers in aviation intermediate-frequency (360 Hz to 800 Hz) power supply systems. Intermediate-frequency vacuum arc experiments were carried out in interrupters with a diameter of 41 mm and CuCr50 alloy contact material. The results show that, as the frequency and peak value of the current increase, both the peak value and rise rate of the intermediate-frequency vacuum arc voltage also increase, and the interruption ability decreases. However, compared to the power frequency current at the same value, the erosion of the contacts is weaker over a shorter arc period. When the vacuum arc reignites, metal droplets are emitted from the contacts. The drive force is from the center of the contact to the edge. If the density of the plasmas and metal vapors and the number of the metal droplets reaches a certain level, the arc may break down, which will cause the interruption to fail.