NASA is leveraging IBM AI for climate change research

ibm and the NASA Marshall Space Flight Center Formed a strategic partnership to leverage IBM’s artificial intelligence technology to analyze geoscience and geospatial data. The amount of data obtained by observation satellites and other sensors is enormous, and to be able to analyze it for correlations it is necessary to use innovative approaches, in this case AI, in order to simplify and speed up the work of researchers.

Specifically, we will use some foundation modelAI models trained on a huge database of unlabeled data. An example of an application of the core model is modern systems that are able to understand natural language and use it to respond.

“Foundational models have proven effective in natural language processing and it is time to extend them to new domains and modalities important to business and society“, he stated Raghu JantiPrincipal researcher at IBM. “The application of baseline models to geospatial data, event sequences, time series and other non-linguistic parameters within Earth science data could suddenly provide information of enormous value to a much larger group of researchers, businesses and citizens. Ultimately, it can make it easier for everyone working on some of the most pressing climate issues“.

NASA and IBM projects

There are currently two projects in the works by IBM and NASA. One of them involves training the algorithm with data collected from Landsat Sentinel-2 Coordinator on changes in the Earth’s crust detected by satellites. Petabytes of information that will help identify changes in the geographic footprint caused by phenomena such as natural disasters, crop rotation patterns, and wildlife habitat patterns. IBM technology will help researchers provide insight into our planet’s ecosystems.

Another project aims to create an accessible collection of Earth-related scientific literature. In this regard, IBM technicians have developed a natural language processing model that will be applied to about 300,000 scientific articles and will allow easy organization and indexing of all information. This model has been trained on Red Hat OpenShift levers PrimeQAan open source, multilingual IBM conversational (question-answer) model training system.

These are not the only projects in the pipeline. IBM and NASA also plan to build base models to make weather forecasts by leveraging the MERRA2 atmospheric observation dataset.

“The good thing about base models is that they can be used for many derivative applications“, he stated Rahul Ramachandran, a senior investigator at NASA’s Marshall Space Flight Center in Huntsville, Alabama. “He added that the creation of these foundational models could not be done by small teams. Teams of different organizations need to contribute their different perspectives, resources and experiences“.