Public Meteorological Service

Powered on the twin engines of high performance compute + AI available with HUAWEI CLOUD, meteorological associations at different levels can share information, converge into omnimedia, enhance user experience of public meteorological service, and push public announcements whenever necessary.
Solution Advantages
  • Superior Performance Experience

    Industry-leading performance with 100G IB compute networking and local 3.2T enterprise-class SSDs (local cache disks available from a cloud service provider).

  • Compatible with Open Source – Beyond Open Source

    Compatible with TensorFlow and MXNet (the two major open source-based deep learning frameworks); dozens of popular CNN/RNN models available including pre-trained models.

  • Superior Value

    Able to synchronize with new GPU technologies, enabling pain-free switchover to the newest hardware. Support for pay-as-you-go and monthly-plan billing models. Order and enjoy, scale-up with ease.

  • Low-Latency Storage Access

    Intelligent scheduling combined with accelerated transmission utility delivers high concurrency and bandwidth while keeping latency low. Users are ensured quick access to the data they need.

Business Challenges
  • Hard to Integrate Data

    Meteorological radar, ground observation, aerial probing, satellite imaging, oceanic surveillance, and other meteorological data systems cannot effectively tie into systems from external sources, taking away from the level of service and accuracy the organizations hope to achieve.

  • Lack of Compute Power

    As the temporal-spatial resolution of various types of observational data increases, the amount of data also increases. Data needs to be more standard and processing capability needs to keep up with demands.

  • Low Forecast Accuracy

    There is plenty of room for improvement in terms of forecast accuracy. The current accuracy in forecasting heavy rain and intense storms is less than 20%. This percentage does not satisfy the need for timely, accurate, and quantitative weather forecasting.

  • Outdated Service Models

    At present, the meteorological service model is relatively simple. Often, customers only get simple values about future weather and simple text-based descriptions.

Typical Scenarios
  • High-Performance Numerical Forecast

  • Intelligent Short-term Forecast

  • Multi-Media Services

High-Performance Numerical Forecast

HUAWEI CLOUD provides high performance bare metal servers and GPU cloud acceleration servers to speed up computations of numerical weather forecasting applications while significantly increasing the execution efficiency.

  1. Flexible Self-Service Capabilities

    Automated provisioning of VMs and cloud-based bare metal, automated cluster creation, automated long-term status detection; importing of various HPC application templates; deployment of MPI libraries and compilation libraries, configuration optimizations, and so on in the VM templates.

  2. Configure Resources You Need to Handle Your Dynamic Workloads

    Near instantaneous build with an almost unlimited infrastructure capacity; flexibility in selecting VM/cloud-based bare metal and various other types of compute/storage instances

  3. High-Performance Network

    Up to 10 GB/s network bandwidth. Single bare metal instances equipped with 100 GB IB networking to best suit the compute cluster requirements for at-volume data transfer

  4. GPU Direct

    Support for the GPU Direct technology to achieve direct connections between units; equipped with NvLink technology to enable a five-fold improvement in data transfer efficiency between GPUs. The technologies help ensure high bandwidth and low latency in data transmissions in addition to robust data processing capabilities.

  5. Elastic Cloud Server

    Auto Scaling

    GPU-accelerated Cloud Server

    Elastic Volume Service

    Virtual Private Cloud

Intelligent Short-term Forecast

A large number of optimized model algorithms are built in the variety of AI services available on the high-performance compute offerings in HUAWEI CLOUD. Accuracy of short-term forecast is improved based on basic radar data and radar product data, using data parsing, precheck information extraction, and check models.

  1. A Variety of Algorithms

    Dozens of CNN/RNN algorithm models such as image classification and object detection available. Large number of trained models based on open-source data sets to help expedite learning.

  2. Efficient Computing

    Accelerated model training with hybrid-parallel, gradient compression, accelerated convolutional neural networks (deep learning), and EASGD technologies. Built-in model compression capabilities significantly reduce model sizing costs.

  3. Ease-of-Use

    Seamlessly interface with HUAWEI CLOUD OBS service and GPU high-performance computing offering to fully satisfy a variety of service requirements.

  4. Professional-grade

    Detection and prediction of extreme weather (hail, rainstorms, high winds) with data from radar pickups.

  5. Elastic Cloud Server

    Deep Learning Service

    Machine Learning Service

    Object Storage Service

Multi-Media Services

Video, image, text, data, and other delivery formats enhance public perception.

  1. Video Playback

    Support for on-demand, live, interactive, and other video playback modes.

  2. Image Search

    Image-tag searches enable users to quickly search for the desired image by typing in keywords or copying the image into the search.

  3. Stable and Reliable

    System supports up to 100 million concurrent connections, fully satisfying at-scale service access needs.

  4. Backup and Recovery

    Up to 35 days of automated system backups before overwrite, PITR to any point in time, support for manual creation of snapshots, quick recovery in downed resource or disaster recovery event.

  5. Massive, Diverse Node Capability

    HUAWEI CLOUD has over 500 acceleration nodes, covering major carriers and small to medium-sized carriers. User requests are sent to the optimal edge node for quick handling.

  6. Elastic Cloud Server

    Content Delivery Network

    Elastic Load Balance

    Object Storage Service

    Distributed Cache Service for Redis

    RDS for MySQL

Recommended Services
  • GPU-accelerated Cloud Server
    GPU-accelerated Cloud Server (GACS) provides outstanding floating-point computing capabilities. They are suitable for scenarios that require real-time, highly concurrent massive computing, such as deep learning, scientific computing, CAE, 3D animation rendering, and CAD.
  • Deep Learning Service
    Deep Learning Service (DLS) is powered on the high-performance computing capabilities of HUAWEI CLOUD. With various built-in neural network models, DLS allows you to easily implement model training, evaluation, and inference all with the flexibility of on-demand scheduling.
  • Content Delivery Network
    Content Delivery Network (CDN) delivers content from origin servers to edge node servers so that users can quickly obtain content from nodes nearest to them. CDN speeds up website response, ensures content availability, and prevents network traffic bottlenecks.

Create an Account and Experience HUAWEI CLOUD for Free

Register Now