A novel discovery concerning the utilization of Artificial Intelligence (AI) has been set to make scientists aware of the accurate forecast of Arctic Sea ice conditions. The amazing predictions can reinforce novel early-alert systems ensuing protection of Arctic wildlife and coastal areas from the negative repercussions of melting sea ice.
British Antarctic Survey and The Alan Turing Institute Described the Efficiency of the AI System
An international team of researchers administered by British Antarctic Survey (BAS) and The Alan Turing Institute expressed the contribution of AI system- IceNet in the journal Nature Communications that how Artificial Intelligence (AI) can accurately forecast the Arctic Sea Ice shrinkage- the task that has been marked as “eluded” since ages. Due to its convoluted association with the atmosphere, the giant layer of sea ice or frozen seawater above the north and south poles is quite challenging to be forecasted.
The sensitivity of sea ice towards rising temperature has led to the shrinkage of Arctic Sea ice over the past few decades. Beyond doubt, these rapid changes dramatically affect the climatic conditions, leading to distinct dangers and health issues to indigenous and local communities.
IceNet- AI-based Tool Boasts the Potency to Predict 95% Accurate Weather
As the task of recording and forecasting Arctic Sea ice shrinkage was treated as “beyond the scope” by the researchers, IceNet, an AI predictive Tool, has created hope amongst all the researchers worldwide. Yes, IceNet can indeed predict weather conditions of Arctic Sea ice with an accuracy of 95%. This discovery has beaten the records of physics-based models as predictions can be made two months ahead. Scientists have announced that IceNet can predict sea ice shrinkage, is a favourable step towards “Arctic Sustainability Efforts,”and works thousands of times more efficiently than conventional methods.
Novel Sea Ice Forecasting System Fuses Data from Satellite Sensors
The researchers have suggested that the new sea ice forecasting mechanism developed with Artificial Intelligence (AI) fuses data from the satellite sensors yielding climate models that conventional systems failed to achieve. In contrast to traditional methods, the newly discovered process does not model the physic laws directly. The AI-based IceNet is designed on a concept called Deep Learning. With the help of this approach, the model assimilates the changes of sea ice dating back to thousands of years of climate simulation data, along with years of observational data to forecast the extent of Arctic Sea ice months into the future.
Researchers Plan to Develop a Daily Version of the AI Model
As the experiment proved successful, researchers are further investigating to achieve their next goal of developing a daily version of the model. They would run the daily version publicly, similar to day-to-day weather forecasts. It would be the most purposeful invention as an early alert system to warn or alert the world for risks associated with sea ice loss.
The Researchers of John Hopkins Also Use Convolutional Neural Networks
Another team of researchers at John Hopkins University Applied Physics Laboratory has launched a forecast model framed using AI, known as “Convolutional Neural Network,”to analyze satellite images of the ocean’s surface while predicting ice formation in the coming weeks. It has been observed that neural networks have the potency to sort digital pixels more promptly than humans, hence being utilized in facial recognition algorithms. The model launched by JHUAPL uses digital satellite images and fuses with meteorological data gathered on the ground.
At present, the US National Ice Center forecasters in Colorado gather weekly forecasts of Arctic Ice by hand while examining the pictures taken by the orbiting satellites and collating them with historical data. Unfortunately, the method is not that beneficial as the Arctic Sea is losing its ice cover rapidly. By 2050, it could be free of ice, as estimated by 21 research institutions in a 2020 publication.
IceNet is Much More Efficient and Faster than SEAS5
IceNet beats the leading physics-based model SEAS5 based on higher efficiency and faster speed. It leaves a charismatic impact. The British Antarctic Survey has devised a much quicker and more efficient AI-based model named IceNet, which runs 2000 times faster than SEAS5 do. SEAS5, a conventional physics-based model, takes about 6 hours on a supercomputer to display a forecast, whereas IceNet can perform the same function in less than 10 seconds on a laptop! Isn’t it a worth appreciating feature that distinguishes it from the rest? The novel discovered system could predict anomalous events happening with ice- surprisingly highs or lows, almost four months in advance, as apprised by the BAS researchers.
IceNetProved 2.9 Per Cent More Accurate Than SEAS5
The conventional toolSEAS5 compared the outputs delivered by IceNet to determine its accuracy level. It was observed that the readings determined by IceNet were 2.9 per cent more accurate as compared to SEAS5, corresponding to a further 360,000 square kilometres of the ocean being accurately tagged as “ice” or “no ice.”IceNet is considered a significant step in forecasting the sea ice shrinkage, amalgamating the ability to deliver accurate forecasting that was not possible through any other conventional methods. Scientists also believe that IceNet has better absorbed the physical processes regulating sea ice growth from the training data, whereas the SEAS5 model still faces difficulty understanding the same information.
SmartIce—Another Innovative Project based on Powered Sensors, Records Ice Thickness
Polarctic’sCanavera is collaborating with officials of Canada to develop forecasts of ice for residents of Nunavut to create a better understanding of the crucial food resource areas. SmartIce is a separate project that utilizes data from small battery-powered sensors embedded in the sea ice that record ice thickness and temperature. The information can be used to assist navigation and safety of the native people. It is indeed an ultimate choice for the natives to get benefitted.
Thus, Artificial Intelligence (AI) and the keen dedication of researchers have constantly contributed to discovering several innovative tools for weather monitoring, climate change, arctic ice records, and many more.