IBM Hackathon Produces Innovative Sustainability Solutions on AWS

This post was co-authored by Dhana Vadivelan, Senior Manager, Solutions Architecture, AWS; Markus Graulich, Chief Architect AI EMEA, IBM Consulting; and Vlad Zamfirescu, Senior Solutions Architect, IBM Consulting

Our consulting partner, IBM, organized the “Sustainability Applied 2021” hackathon in September 2021. This three-day event aimed to generate new ideas, create reference architecture patterns using AWS Cloud services, and produce sustainable solutions.

This post highlights four reference architectures that were developed during the hackathon. These architectures show you how IBM hack teams used AWS services to resolve real-world sustainability challenges. We hope these architectures inspire you to help address other sustainability challenges with your own solutions.

IBM sustainability hackathon

Any sustainability research requires acquiring and analyzing large sustainability datasets, such as weather, climate, water, agriculture, and satellite imagery. This requires large storage capacity and often takes a considerable amount of time to complete.

The Amazon Sustainability Data Initiative (ASDI) provides innovators and researchers with sustainability datasets and tools to develop solutions. These datasets are publicly available to anyone on the Registry of Open Data on AWS.

In the hackathon, IBM hack teams used these sample datasets and open APIs from The Weather Company (an IBM business) to develop solutions for sustainability challenges.

Architecture 1: Biocapacity digital twin solution

The hack team “Biohackers” addressed biocapacity, which refers to a biologically productive area’s (cropland, forest, and fishing grounds) ability to generate renewable resources and to absorb its spillover waste. Countries consume 1.7 global hectares per person of Earth’s total biocapacity; this means we consume more resources than are available on this planet.

To help solve this problem, the team created a digital twin architecture (Figure 1) to virtually represent biocapacity information in various conditions:

• The Satellite Image Rekognition Module (SIRM) receives biocapacity data from the sustainability category of the Registry of Open Data on AWS datasets and customer on-premises servers via AWS Storage Gateway. SIRM also receives ground data from satellite images, which are processed using Amazon Rekognition, and the data is stored in Amazon Simple Storage Service (Amazon S3).
• The Biocapacity Intelligent Forecasting Module (BIFM) combines biocapacity data and extracted satellite images using AWS Glue. Amazon Forecast predicts biocapacity risk information like fishing or built-up land.
• The forecast optimization module (FOM) analyzes the biocapacity risks and shows possible actions that can be taken to improve. Amazon Elastic Compute Cloud (Amazon EC2) provides a virtual representation of the current and future biocapacity information to users using Amazon CloudFront.

Figure 1. Biocapacity digital twin solution on AWS

This solution provides precise information on current biocapacity and predicts the future biocapacity status at country, region, and city levels. The Biohackers aim to help governments, businesses, and individuals identify options to improve the supply of available renewable resources and increase awareness of the supply and demand.

Architecture 2: Local emergency communication system

Team “Sustainable Force 99,” worked to better understand natural disasters, which have tripled over recent years. Their solution aims to predict natural disasters by analyzing the geographical spread of contributing factors like extreme weather due to climate change, wildfires, and deforestation. Then it communicates clear instructions to people who are potentially in danger.

Their solution (Figure 2) uses the Amazon SageMaker AI prediction model and combines The Weather Company data with Amazon S3, Amazon DynamoDB, and Serverless on AWS technologies and communicates with the public by:

• Providing voice and chat communication channels using Amazon Connect, Amazon Lex, and Amazon Pinpoint.
• Broadcasting alerts across smartphones and speakers in towns to provide information to people about upcoming natural disasters.
• Allowing individuals to notify emergency services of their circumstances.

Figure 2. Local emergency communication system

This communication helps governments and local authorities identify and communicate safe relocation options during disasters and connect with nearby relocation assistance teams.

Architecture 3: Intelligent satellite image processing

Team “Green Fire” examined how to reduce environmental problems caused by wildfires.

Many forest service organizations already use predictive tools to provide intelligence to fire services authorities using satellite imaging. The satellites can detect and capture images of wildfires. However, it can take hours to process these images and transmit and analyze the information, which can mean the subsequent coordination and mobilization of emergency services can be challenging.

To help improve fire rescue and other emergency services coordination and real-time situational awareness, Green Fire created a serverless application architecture (Figure 3):

• The drone images, external data feeds from AWS Open Data Registry, and the datasets from The Weather Company are stored in Amazon S3 for processing.
• AWS Lambda processes the images and resizes the data, and the outputs are stored in DynamoDB.
• Rekognition labels the drone images that identify geographical risks and fire hazards. This allows emergency services to identify potential risks, and they can further model the potential spread of the wildfire and identify options to suppress it.
• Lambda integrates the API layer between the web user interface and DynamoDB so users can access the application from a web browser/mobile device (mobile application developed using AWS Amplify). Amazon Cognito authenticates users and controls user access.

Figure 3. Intelligent image processing

The team expects this solution to help prevent loss of human lives, save approximately 10% of yearly spend, and reduce CO2 emissions globally between 1.75 to 13.5 billion metric tons of carbon.

Architecture 4: Machine-learning-based solution to predict bee colonies at risk

Variations in vegetation, rising temperatures, and use of pesticide are significantly destroying habitats and creating inhospitable conditions for many species of bees. Over the last decade, beehives in the US and Europe have suffered hive losses of at least 30%.

To encourage the adoption of sustainable agriculture, it is important to safeguard bee colonies. The “I BEE GREEN” team developed a device that contains sensors that monitor beehive health, such as temperature, humidity, air pollutants, beehive acoustic signals, and the number of bees:

• AWS IoT Core for LoRaWAN connects wireless sensor devices that use the LoRaWAN protocol for low-power, long-range comprehensive area network connectivity.
• The sensor and weather data (from AWS Open Data Registry/The Weather Company) are loaded in Amazon S3 and transformed using AWS Glue.
• Once the data is processed and ready to be consumed, it is stored on the S3 bucket.
• SageMaker consumes this data and generates forecasts such as estimated bee population and mortality rates, probability of infectious diseases and their spread, and the effect of the environment on beehive health.
• The forecasts are presented in a dashboard using AWS QuickSight and are based on historical data, current data, and trends.

The information this solution generates can be used for several use cases, including:

• Offering agricultural organizations information on how climate change impacts bee colonies and influences their business performance
• Contributing to beekeepers’ understanding of the best environment to safeguard bee biodiversity and when to pollinate crops
• Educating farmers on sustainable methods of agriculture that reduces threat to bees

The I BEE GREEN device automatically counts when bees are entering or exiting the hive. Using this, farmers can assess a hive’s acoustic signal, temperature, humidity, the presence of pollutants, classify them as normal or abnormal, and forecast the potential risk of abnormal mortality. The I BEE GREEN team aims for their device to deploy quickly, with minimal instructions anyone can implement.

Figure 4. Machine-learning-based architecture to predict bee colonies at risk

Conclusion

In this blog post, we showed you how IBM hackathon teams used AWS technologies to solve sustainability-related challenges.

Ready to make your own? Visit our sustainability page to learn more about AWS services and features that encourage innovation to solve real-world sustainability challenges. Read more sustainability use cases on the AWS Public Sector Blog.