Abstract Correcting the forecast bias of numerical weather prediction models is important for severe warnings. The refined grid requires direct correction on gridded products, as opposed to correcting data only at individual stations. In this study, a deep learning method called CU-net proposed correct forecasts four variables from European Centre Medium-Range Weather Forecast Integrated Forecasting System global model (ECMWF-IFS): 2-m temperature, relative humidity, 10-m wind speed, and...

Iron ore feed-load control is one of the most critical settings in a mineral grinding process. It has direct impact on quality final iron products. The setting feed load heavily replies characteristics pellets. However, such are challenging to acquire many production environments, requiring speical equipments and complicated modelling process with high cost. To provide an low-cost easier-to-implement solution, this paper, we present our work using deep learning models for estimation from...

Abstract This study investigates the roles of atmospheric moisture transport under influence topography for summer extreme precipitation over North China (NC) during 1979–2016. Based on rain gauge data and a reanalysis, 38 days in NC years were selected associated fluxes estimated. The results show that there is dominant influx 311.8 kg m −1 s into along its southern boundary from tropical oceans, secondary 107.9 across western carried by mid-latitude westerlies. outflux eastern 206.9...

As the resolution of Earth observation imagery advances, many downstream industries that rely on remotely sensed can also advance. Due to physical limitations optical sensors carried by satellites, growth remote sensing images becomes difficult. Therefore, it is becoming increasingly important boost methods other than upgrading components, such as super-resolution. a computer vision task, convolutional neural networks (CNNs) perform well super-resolution tasks. Transformer-based models show...

Modern radar systems have high requirements in terms of accuracy, robustness and real-time capability when operating on increasingly complex electromagnetic environments. Traditional signal processing (RSP) methods shown some limitations meeting such requirements, particularly matters target classification. With the rapid development machine learning (ML), especially deep learning, researchers started integrating these new solving RSP-related problems. This paper aims at helping...

Tobacco is a critical cash crop in China, so its growing status has received more and attention. How to acquire accurate plant area, row spacing, spacing at the same time have been key points for grow monitoring yield prediction. However, accurately detecting small densely arranged tobacco plants during rosette stage poses significant challenge. In Sichuan Province, contours of scattered fields with different shapes are not well-extracted. Additionally, there lack simultaneous methods...

Abstract Using observation data and the simulations of Variational Doppler Radar Analysis System (VDRAS), convection initiation (CI) ahead a gust front over Yangtze River Delta in Jiangsu province, China on 5 June 2009 was studied. The CI located exit region upper level jet with moderate CAPE. It is found that, there northeast‐southwest convergence line near ground front, which could not trigger new alone. upper‐level updraft above surface alone either. superposition low levels promoted...

In recent years, large-scale pre-trained multimodal models (LMM) generally emerge to integrate the vision and language modalities, achieving considerable success in various natural processing computer tasks. The growing size of LMMs, however, results a significant computational cost for fine-tuning these downstream Hence, prompt-based interaction strategy is studied align modalities more efficiently. this contex, we propose novel inspired by human memory strategy, namely Memory-Inspired...

With the rapid development of augmented reality (AR) and virtual (VR) technology, human-computer interaction (HCI) has been greatly improved for gaming AR VR control. The finger micro-gesture is a hot research focus due to growth Internet Things (IoT) wearable technologies recently Google developed radar based sensor which Soli. Also, there are number techniques have using Time Flight (ToF) imaging sensors 3D glasses such as Atheer mobile glasses. principle holoscopic (H3D) mimics fly's eye...

The trend, severity, and duration of drought in the eastern fringe Tibetan Plateau (EFTP) have been investigated using Mann-Kendall (M-K) trend test, standardized precipitation index (SPI), generalized extreme value (GEV), data obtained from 438 rainfall stations reanalysis datasets for period 1961–2014. A recent is evident a decrease rainfall, with this mainly occurring on slope TP (< 3000-m elevation); attributed to downward air flows over (TP) induced by heating. Recent droughts also more...

Abstract The ability to forecast thermodynamic conditions aloft and near the surface is critical accurate forecasting of precipitation type at surface. This paper presents an experimental version a new scheme for diagnosing type. method considers optimum temperature threshold associated with each combines model-based explicit fields hydrometeors higher-resolution modified topographic information determine types in North China. Based on over 60 years precipitation-type samples from China,...

A case study is presented of convection initiation (CI) resulting from the merger a cold front with dryline in southwestern Beijing, China, on afternoon 11 June 2011. This process analyzed S-band Doppler radar data, surface automatic weather station and mesoscale numerical simulation results. The formation this analogous to that Great Plains United States, it conducive CI updrafts generated baroclinic frontogenesis, favorable instability immediately moist side. Prior front–dryline merger, as...

Personal Credit Scoring is of great significance for commercial banks to circumvent credit consumption, the original BP algorithm's convergence rate slow, learning precision low, training process easy fall into local minimum, this paper presents an improved algorithm with variable based on algorithm, and applied simulate personal scoring. After comparing we found has greatly reduced network's number iterations, shorten network time 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 Radar Analysis System assimilation analysis. During the process, gust front appeared ahead two existing systems, respectively. In warm moist environment fronts south, there was mesoscale air mass boundary. With process moving southward, approaching...

With the advent of era big data, personalized recommendation is becoming more and important. However, there a problem data sparsity in recommender system. The sparseness leads to inaccuracy recommendation. rapid development social network, researchers begin add relationship into Trust very important kind relationships. In this paper, we bring up new approach for rating prediction by using trust computation. First, describe method Then, give strategy user's preference. Finally, compare our...

Tobacco is one of the most important cash crops in China, and accurate acquisition its planting information great significance for tobacco management yield prediction. In this paper, a extraction method based on UAV high-resolution images proposed. The uses RtinaNet target detection model to achieve identification localization images, generates coordinate information. Based coordinates, we designed two algorithms extracting plant spacing, row spacing area fields, respectively. experimental...

Blind super-resolution (BSR) methods based on high-resolution (HR) reconstruction codebooks have achieved promising results in recent years. However, we find that a codebook HR may not effectively capture the complex correlations between low-resolution (LR) and images. In detail, multiple images produce similar LR versions due to blind degradations, causing HR-dependent only having limited texture diversity when faced with confusing inputs. To alleviate this problem, propose Rich...

With the rapid development of augmented reality (AR) and virtual (VR) technology, human-computer interaction (HCI) has been greatly improved for gaming AR VR control. The finger micro-gesture is one important interactive methods HCI applications such as in Google Soli Microsoft Kinect projects. However, progress this research slow due to lack high quality public available database. In paper, holoscopic 3D camera used capture images a new unique (HoMG) database produced. principle based on...

Augmented reality (AR) technologies provide a novel experience for users to interact with the real-world environment by integrating virtual content into real circumstances. However, wireless AR applications have trouble ensure visual security in complex and various scenarios. In this paper, we propose reinforcement learning based AR, which adaptively revises location of avoid obscuring context. The devices make display strategies according partially observable Markov decision procession...

&amp;lt;p&amp;gt;The objective extrapolation forecast is the main method for 0-1 hour convective storm nowcasting. Radar echo was performed by using 6 minute interval radar mosaics obtained from images of 8 multi-radars in Beijing-Tianjin-Hebei region. A comparative study two extrapolated forecasts eighteen typical precipitation events occurred region 2016 to 2018 conducted. Compared with tracking echoes correlation method, variational utilizes technique compute motion vector fields, and...