AI-Powered Solar Forecasting Model Integrates Indigenous Calendars for Enhanced Accuracy

Indigenous calendars could make solar power more efficient

A truly sustainable future requires solar power, but trying to consistently maximize the energy harvested by panel arrays remains one of the industry's biggest challenges. Unlike fossil fuels, solar power yields are dictated by the complex interplay of weather and atmospheric variables, as well as the sun's own activity. This means it's basically impossible to craft a universal prediction model, so localized solar forecast systems are a necessity. While machine learning technology has significantly improved today's forecast models, there is still a lot of room for improvement. But an artificial intelligence program is only as good as the data used to train it -- and according to researchers at Australia's Charles Darwin University, it's tough to find a better solar forecasting dataset than First Nation seasonal calendars. Their new approach is detailed in a study published in the IEEE Open Journal of the Computer Society. Present-day non-Indigenous cultures generally divide the year into four seasons, but that's not the case for many past and present Indigenous communities. Solar calendars like the Aztecs' were accurate enough to guide farming practices that fed millions of people, for example. In Australia, the people of the Tiwi Islands use a three-season calendar based on their local ecological knowledge. Darwin's Gulumoerrgin (Larrakia) community recognizes seven principle seasons, while the Kunbarlanja (Gunbalanya) and Ngurrungurrudjba of the Northern Territory also possess nuanced calendars of their own. "These calendars are closely tied to weather patterns and seasons. The deep understanding of local climate in these calendars enables First Nations people to make informed resource management and sustainability decisions," the study's authors explain. "As climate change affects weather patterns, knowledge of these calendars becomes crucial for adapting to environmental challenges." Additionally, unlike conventional calendar systems, Australia's Indigenous cultures base their seasonal classifications on local ecological cues. These include plant and animal behaviors that closely relate to shifting sunlight and weather patterns. With this in mind, the team broke down information into various datapoints from the Tiwi, Gulumoerrgin (Larrakia), Kunwinjku, and Ngurrungurrudjba First Nations calendars, along with a modern calendar known as Red Centre. Researchers then entered their First Nations Seasonal Metrics (FNS-Metrics) dataset into a novel machine learning model designed to detect large-scale patterns. From there, they tested the system against past solar power and weather information collected by the Desert Knowledge Australia Solar Centre (DKASC) in Alice Springs. The results were striking: the First Nations Seasonal Metrics vastly outperformed many of today's leading forecasting models. Compared to an already strong baseline model, the First Nations data were 14.6 percent more accurate while reducing the error rate by 26.2 percent -- less than half the error rate of existing forecasting programs. "Incorporating First Nations seasonal knowledge into solar power generation predictions can significantly enhance accuracy by aligning forecasts with natural cycles that have been observed and understood for thousands of years," said Luke Hamlin, a CDU Ph.D candidate and study co-author who is also a member of eastern Australia's Bundjulang nation. Hamlin added that integrating Indigenous knowledge into predictive models can more closely tailor a system to reflect the more nuanced shifts in environmental conditions. This offers "more precise and culturally informed forecasting" for individual regions. The team says this strategy is also particularly promising for rural communities already home to larger First Nation communities. These same places could benefit the most from additional solar farms. And the approach isn't just limited to solar power. "In future work we'll explore the applications of the model to other regions and renewable energy sources," said Thuseethan Selvarajah, a CDU information technology lecturer and study co-author.

World-first study uses First Nations calendars for solar power forecasting

The in-depth observations of First Nations seasonal calendars could be key to improving solar power forecasting, according to a world-first study by Charles Darwin University (CDU). The study, "Conv-Ensemble for Solar Power Prediction with First Nations Seasonal Information" published in IEEE Open Journal of the Computer Society, combined First Nations seasonal calendars with a novel deep learning model, an artificial intelligence technique, to predict future solar panel power output. Solar is one of the world's leading renewable energy alternatives but there continue to be challenges with the technology's reliability. At present, solar power generation is difficult to predict because of weather, atmospheric conditions and how much power is absorbed on a panel surface. CDU researchers developed the model using the Tiwi, Gulumoerrgin (Larrakia), Kunwinjku and Ngurrungurrudjba First Nations calendars, and a modern calendar known as Red Center. Researchers used data from the Desert Knowledge Australia Solar Center in Alice Springs, and the results show the model can predict solar power generation with a lower error rate. The error rate is less than half of the error rate that popular forecasting models use in the industry right now. Co-author, CDU Ph.D. student and Bundjalang man Luke Hamlin said the environmental knowledge held within these calendars was an invaluable resource. "Incorporating First Nations seasonal knowledge into solar power generation predictions can significantly enhance accuracy by aligning forecasts with natural cycles that have been observed and understood for thousands of years," Mr. Hamlin said, "Unlike conventional calendar systems, these seasonal insights are deeply rooted in local ecological cues, such as plant and animal behaviors, which are closely tied to changes in sunlight and weather patterns. "By integrating this knowledge, predictions can be tailored to reflect more granular shifts in environmental conditions, leading to more precise and culturally informed forecasting for specific regions across Australia." Associate Professor in Information Technology Bharanidharan Shanmugam and Lecturer in Information Technology Dr. Thuseethan Selvarajah, who are co-authors of this paper, said the combination of advanced artificial intelligence and ancient First Nations wisdom could revolutionize prediction technology. "Accurate solar power prediction is challenging, and these challenges hinder the development of a universal prediction model," Associate Professor Shanmugam said. "The success of the proposed approach suggests that it could be a valuable tool for advancing solar power generation prediction in rural areas, and in future work we'll explore the applications of the model to other regions and renewable energy sources," Dr. Selvarajah said.