Improving weather forecasting accuracy using Amazon AppStream 2.0 graphic-intensive instances

The ability to access applications and data from anywhere on any device is increasingly important for remote, disconnected, and distributed workers. In-field researchers and meteorologists throughout the globe depend on real-time data ingestion for essential weather capabilities.

The Advanced Weather Interactive Processing System (AWIPS) is a weather forecasting and display system used by the National Weather Service (NWS) and other meteorological agencies worldwide. The system allows meteorologists to monitor, analyze, and predict weather patterns with greater accuracy, resulting in better forecasts and more effective warnings to help protect lives and property. AWIPS integrates data from various sources such as satellites, radar, and weather balloons. The Common AWIPS Visualization Environment (CAVE) is the client-side application that forecasters use to interact with the AWIPS system. CAVE receives processed data from AWIPS servers and provides the visualization tools that let meteorologists create forecasts, issue warnings, and visualize weather data.

In this blog post, we will show you how to stream the Common AWIPS Visualization Environment (CAVE) application using Amazon AppStream 2.0—a managed application streaming service that enables users to stream desktop applications from Amazon Web Services (AWS) to any device without additional hardware or software.

Benefits of running graphic-intensive applications using Amazon AppStream 2.0

AppStream 2.0 provides numerous benefits. With this streaming service:

• Users can access their desktop applications from any supported device, including laptops, desktops, tablets, and phones. This allows them flexibility and mobility, enabling employees to remain productive whether they’re in the office, working from home, or traveling.
• You can easily scale your GPU capacity up or down based on demand. This allows you to optimize costs by only paying for what you use, eliminating the need to maintain expensive hardware that may sit idle during off-peak hours.
• You don’t have to overprovision GPU resources, which is particularly valuable for organizations with variable workloads or seasonal demands.
• Users can access GPU-accelerated applications from any device, including mobile. This allows more flexibility than having the applications tied to specific desktops, and provides consistent performance regardless of the end user’s device capabilities.
• AppStream 2.0 handles the complexities of managing GPU drivers, GPU-enabled fleets, and GPU-optimized application stacks. This lowers the administrative burden on your IT staff, freeing them to focus on strategic initiatives rather than routine maintenance and troubleshooting.
• The GPUs are provided on a pay-as-you-go model in AppStream 2.0, the high upfront costs of buying and maintaining physical GPU servers are eliminated. This makes it more affordable for many organizations, particularly those just starting out or looking to expand their graphics capabilities without major capital expenditure.
• AppStream 2.0 offers instance types optimized for graphics workloads like Graphics G4dn and Graphics G5 instances powered by NVIDIA GPUs. This provides high-performance for tasks like 3D modeling and video editing, and gives you the ability to choose the most appropriate instance type for specific application requirements and user needs.
• Your GPU workloads and data never reside on end-user devices and are isolated in the AWS Cloud. This enhances security and compliance for sensitive applications, while providing centralized control over data access and reducing the risk of intellectual property loss through local device compromise.

Figure 1. AWIPS CAVE Visualization of Hurricane Lee September 13, 2023 using Amazon AppStream 2.0. 

Exploring the AWIPS solution

The AWIPS CAVE workstation is the primary interface used by forecasters. AWIPS Cave has the following system requirements:

• OpenGL 2.0 Compatible Devices
• At least 4GB RAM
• At least 2GB Disk Space for Caching
• NVIDIA Graphics Card
• Latest NVIDIA Driver

To configure the AWIPS application for Amazon AppStream 2.0, you’ll perform the following steps:

Create Linux Based Image

First, we’ll prepare the base Linux environment and install AWIPS CAVE with its dependencies. This creates the foundation for our streaming application.

Create Application Optimization Manifest

Amazon AppStream 2.0 uses manifest files to optimize application streaming performance by pre-loading necessary files. This script helps identify all files needed by CAVE during runtime.

Add Application to Amazon AppStream Catalog

Register CAVE in the Amazon AppStream application catalog, specifying its launch path, display name, and the optimization manifest we created.

Configure Fleet

Create a fleet to manage the instances that will run your streaming applications. Choose between on-demand (cost-effective) or always-on (instant availability) configurations.

• It is recommended that you choose Graphics G4dn (stream.graphics.g4dn.xlarge) or Graphics Pro (stream.graphics-pro.4xlarge)
• Select “On-demand” fleet type
• Set desired capacity limits

Create Image

Create the final Amazon AppStream image that will be used to launch streaming sessions. This process captures all installed applications and configurations.

Configure User Authentication

Choose between managing users directly in Amazon AppStream or integrating with your existing identity provider.

• Configure Amazon Cognito User Pool Authentication
• Integrate Amazon AppStream Environment with AWS Identity Center.

The following figure shows the high-level architecture of the solution.

Figure 2. Architectural diagram of the solution described in this blog. The major components of the solution are Amazon AppStream 2.0, Amazon S3, Amazon EFS, Amazon Aurora, and Amazon EC2.

Field use cases for AWIPS

NWS River Forecast Centers (RFCs) use AWIPS CAVE to monitor and predict river and stream levels, issue flood warnings, and provide information on water supply and drought conditions. Weather Forecast Offices (WFOs) rely on AWIPS CAVE to provide the public with localized weather forecasts, warnings, and advisories. These offices also use AWIPS CAVE to collaborate with other WFOs, federal and state agencies, and private-sector partners. The NWS National Hurricane Center (NHC) uses AWIPS CAVE to monitor and forecast tropical cyclones, including hurricanes and tropical storms. AWIPS CAVE assists the NHC in providing timely and accurate information to the public, emergency management officials, and other stakeholders. Additionally, some universities and research institutions focusing on meteorology and atmospheric sciences also use AWIPS CAVE for research, teaching, and training purposes.

Conclusion

The integration of AWIPS CAVE with Amazon AppStream 2.0 represents a significant advancement in weather forecasting technology and accessibility. This solution addresses the growing need for remote access to critical applications in the meteorological field, while also providing numerous benefits in terms of flexibility, cost-efficiency, and performance.

By leveraging the power of cloud computing and GPU acceleration, meteorologists and researchers can now access sophisticated weather analysis tools from anywhere, on any device. This not only enhances their ability to provide accurate and timely forecasts but also improves the overall efficiency of weather services.

As we continue to face increasingly complex weather patterns and climate challenges, such technological innovations will play a crucial role in supporting the work of meteorologists and ultimately contributing to public safety and preparedness.