AI-based research provides insights into Antarctic moss and climate change

AI technology at the forefront of climate research

Antarctica’s role in regulating Earth’s climate makes it a center for climate research. Most research focuses on carbon uptake in the Southern Ocean or the continent’s reflecting glaciers, but a team of Australian scientists is exploring another aspect: the health of Antarctic mosses. Leveraging an AI-powered edge computing platform, these researchers aim to understand how warming Antarctica could affect global climate patterns.

Importance of Mosses in Antarctic Ecosystems

Mosses cover less than 1% of Antarctica’s surface, but their ecological importance is significant. Moss absorbs atmospheric CO₂ It provides a habitat for important microorganisms, fungi and microscopic animals that make up an important part of the Antarctic food chain. These organisms have adapted to survive in the extreme conditions of Antarctic winters through desiccation and freezing. Over the past two decades, scientists have observed worrying declines in moss health due to changing wind patterns and ozone layer depletion due to climate change. This degradation threatens soil stabilization, CO .₂ Containment and maintaining biodiversity.

Innovative monitoring through AIoT platform

To address these challenges, researchers at SAEF (Securing the Future of the Antarctic Environment) developed a year-round autonomous monitoring system. This system, called an artificial intelligence of things (AIoT) platform, is equipped with solar panels and insulated batteries and can operate even in the harsh Antarctic environment. Built on the NVIDIA Jetson Orin Nano, this platform is designed to efficiently handle AI workloads. It leverages an image segmentation model trained on NVIDIA A100 Tensor Core GPUs. The AIoT platform collects data on moss canopy, temperature, humidity, soil moisture and more and analyzes it at the edge to reduce data being sent back to Australia.

Optimize data processing and transmission

Using NVIDIA TensorRT, image segmentation models are optimized to improve processing speed and reduce power consumption, according to SAEF’s Professor Sharon Robinson. These improvements extend the battery life of AIoT platforms, enabling continuous monitoring. The platform processes data locally, transmitting only essential insights via the LoRaWAN protocol to the Antarctic Research Station and then to the SAEF database. This approach minimizes bandwidth usage, enabling real-time environmental monitoring and analysis.

Future prospects and cooperative efforts

SAEF’s research is supported by NVIDIA’s academic grant program, which provides the necessary computing resources. The team is also working with the Australian Antarctic Department on the Antarctic Ground and Offshore Observation System project, integrating the Jetson platform into a new remote sensing tower. The goal of this collaboration is to improve the interoperability of remote monitoring systems and expand the scope of environmental data collection across Antarctica. For more information, see (NVIDIA Technology Blog)(https://developer.nvidia.com/blog/ai-investigates-antarcticas-disappearing-moss-to-uncover-climate-change-clues/).

Image source: Shutterstock