JAKARTA - The National Research and Innovation Agency (BRIN) emphasizes the importance of using satellites to predict weather changes. This was conveyed in a guest lecture with Tohoku University.
Head of the Climate and Atmosphere Research Center of BRIN, Albertus Sulaiman, said that satellite technology is crucial for maritime and tropical countries, including Indonesia. Moreover, if the data generated is processed with a physics and Artificial Intelligence (AI) approach.
"We see that the integration of the approach of physics and AI in satellite data processing is the future direction of atmospheric research. Indonesia as a maritime and tropical country needs fast and accurate weather monitoring technology," said Albertus, quoted on Friday, February 27.
Satellite data analysis is also closely related to hydrometeorological disaster early warning systems such as floods, landslides, and others. Therefore, BRIN believes that the capacity of satellite data analysis needs to be strengthened.
Meanwhile, Professor Hironobu Iwabuchi from Tohoku University explained that Japan's Himawari-8 and Himawari-9 satellites are able to produce very accurate radiation data for scientists. This is because this satellite is equipped with infrared rays.
The resulting data includes brightness levels, emission temperatures, and spatial textures of cloud patterns in the atmosphere. Scientists can then use complex calculations of the data to predict the coming weather.
However, the purely physics-based method turns out to take a very long time in the processing process. To analyze a full disk containing 36 million pixels, it takes about 100 hours using 10 CPU units.
As a solution, AI-based Convolutional Neural Network (CNN) technology is introduced to speed up the process. Unlike the physics method that calculates per particle, AI works by recognizing patterns of images and cloud textures.
"The process CNN needs to provide predictions based on the shape or pattern of cloud changes only takes 40 seconds. So, this is a rapid advance from previous methods," explained Hironobu.
The English, Chinese, Japanese, Arabic, and French versions are automatically generated by the AI. So there may still be inaccuracies in translating, please always see Indonesian as our main language. (system supported by DigitalSiber.id)
