[Report] Digital Breeding Using Supercomputer Analysis of 120 Days of Precise Rice and Soybean Imaging to Identify Drought and Pest-Resistant Varieties

Rice and soybean plants planted in large pots inside a greenhouse automatically move on a conveyor belt every two days for video recording. This process is to collect big data on 100 different varieties of rice and soybeans, each with distinct genomic information.

On the 5th, we visited the Crop Phenotype Research Building at the National Institute of Agricultural Sciences, Rural Development Administration, located in Jeonju. This facility was completed in 2017 to enable high-speed, large-scale analysis of crop traits. It is the largest research facility in Korea, equipped with visible light, near-infrared, and fluorescence sensors integrated with conveyor systems and robotics automation equipment, capable of capturing and analyzing up to 1,012 individual plants. Based on big data such as genomic information of each variety, this is where 'digital breeding' begins, utilizing artificial intelligence (AI) to shorten breeding periods and increase efficiency.

A rice variety growing in the Crop Expression Research Building of the National Institute of Agricultural Sciences, Rural Development Administration, located in Jeonju.

Upon entering the smart greenhouse, 100 varieties of rice and soybeans with different genetic information growing in separate pots caught the eye. These crops grow here for 120 days, and phenotypes are collected through precise imaging every two days. Phenotypes are the visible growth characteristics of crops, expressed through the interaction between genes and the cultivation environment. Even varieties with the same genes show different growth depending on the environment, and these results are the phenotypes. In the phenotype research building, varieties with different genotypes are grown under the same environment to secure phenotype data.

Kim Kyunghwan, an agricultural researcher at the Agricultural Science Institute, explained, "We use a conveyor belt to move the plants every two days for precise video recording. By controlling the environment uniformly, we measure and record the growth, pot weight, and water amount of each plant with different genetic information, and convert this into data."

Soybean varieties growing in the Crop Expression Research Building are being transported on a conveyor belt for precise imaging.

The big data collected in the phenotype research building is sent to the Agricultural Life Supercomputing Center for analysis. The Rural Development Administration invested a total of 14.8 billion KRW to build the Agricultural Life Supercomputing Center, a two-story above-ground and one-story underground facility covering 2,057 square meters, which was completed last September. Along with the completion, they took over management of the Korea Meteorological Administration's Supercomputer No. 4 and introduced Supercomputer No. 2 to support digital breeding, agricultural meteorology, pest and disease research, and various interdisciplinary studies and shared use.

The advantage of the supercomputer is its ability to analyze data very quickly. For example, analyzing the genotypes of 3,024 rice resources takes six months on a regular computer but only four days on the supercomputer. Lee Taeho, head of the High-Performance Computing Center at the Agricultural Science Institute, said, "Digital breeding differs from traditional breeding methods by digitizing various information obtained from agricultural fields and using big data and AI technology to shorten the breeding period. Using Supercomputer No. 2, we have established a foundation to utilize digital big data in breeding by rapidly analyzing genotype data of economically important crops such as chili peppers, soybeans, and rice, which have high demand for analysis from seed companies and others."

On the 5th, Lee Tae-ho, Director of the High-Performance Computing Center at the Agricultural Science Institute of the Rural Development Administration, explained the supercomputer.

The Rural Development Administration is also promoting the development of agricultural satellites for more accurate agricultural observation. Currently, agricultural observation information is produced by visiting fields to conduct surveys, interviews, and on-site investigations. However, using satellites can produce objective and timely information, increasing efficiency. To this end, the Rural Development Administration is developing the 4th next-generation medium-sized satellite for wide-area agricultural and forestry observation in collaboration with the Korea Aerospace Research Institute and the Korea Forest Service.

An official from the Rural Development Administration said, "Using the agricultural and forestry satellite scheduled for launch in the second half of 2025, we aim to establish technology and observation systems to estimate the area and growth of major crops such as rice, soybeans, onions, and garlic to support agricultural supply and demand policies. We are preparing crop condition prediction and observation systems for immediate use after the satellite launch, and plan to expand the target crops and improve accuracy further through multi-image fusion utilization."