IBM recently launched its most powerful geospatial foundational model, helping at-risk communities better understand and prepare for the effects of the climate crisis, according to TechRadar.
With Prithvi-EO-2.0 featuring 600 million parameters, the new artificial intelligence is trained on NASA's HLS V2 product, which allows for 30-meter granularity. The model is able to find individual tree species and crops using satellite images while also using drone footage to spot cattle and solar panels, per TechRadar.
Juan Bernabe-Moreno, director of IBM Research Europe and Ireland & UK, shared that the model is trained on two types of data: Harmonized Landsat Sentinel-2 (HLS) and Modern-Era Retrospective analysis for Research and Applications (MERRA-2).
HLS data enables users to improve models specializing in the detection of wildfires, flooding, and landslides as well as those capable of measuring canopy heights and tracking land usage.
The other data model, MERRA 2, can track weather — including hurricanes and turbulence — in addition to being capable of weather modeling and temperature mapping.
Countries that previously didn't have access to such resources can now use these tools to plan for the effects of rising global temperatures on their communities, care for their ecosystems, and warn populations ahead of dangerous climate-related events.
Prithvi 2.0 has also expanded on its Multi-Temporal Cloud Gap Imputation that enables the model to recreate areas covered by clouds, which helps improve accuracy with crop yield estimations, planning for agricultural changes, and analyzing environments.
IBM began a partnership with the government of Kenya in 2023 with the goal of supporting their reforestation efforts. With the addition of this new tech, IBM has also been able to assist them with obtaining funding for their reforestation work and creating a flood warning system for at-risk communities.
"The Global South is at the forefront of climate change," said Ambassador Philip Thigo, special envoy on technology for the Republic of Kenya, according to TechRadar.
|
Will America someday get all its energy from renewable sources? Click your choice to see results and speak your mind. |
Yes — very soon ☀️ Geospatial AI has also been useful in detecting variations of heat in African urban areas, providing the capability of modeling health risks based on heat islands and other climate issues.
The long-term goal for IBM is to use what it learned from its development of Prithvi 2.0 to build an open-source model, GridFM, that can oversee power infrastructure and renewable energy.
It will be capable of predicting clean energy generation and tracking transmission of power, helping to forecast energy demands and costs.
"By 2050, power production will need to double in order to meet decarbonization goals," said François Mirallès, a researcher at Hydro Québec, per TechRadar. GridFM could help facilitate the clean energy transition.
Join our free newsletter for weekly updates on the latest innovations improving our lives and shaping our future, and don't miss this cool list of easy ways to help yourself while helping the planet.
Cool Picks
