On September 14, in Yongfeng Village, Taihe Town, Dongpo District, Meishan City, Sichuan Province, harvesters moved back and forth, completing rice cutting, threshing, and straw chopping in one continuous flow. Beside the field ridges, agricultural experts and growers gathered around the yield measurement site. The Sichuan Provincial Department of Science and Technology organized an expert group to conduct on-site yield measurement, and 298 mu of demonstration fields delivered their result: the measured average rice yield per mu reached 863.6 kilograms.

Over three years, the per-mu yield rose from 711 kilograms to 863.6 kilograms, an increase of 152.6 kilograms. Behind these numbers lies the deep cultivation of improved seeds and methods, and more importantly, a shift from "relying on the weather" to "farming by data." As the place where the call to build a higher-level "Tianfu Granary" in the new era was first proposed, on the fertile fields of Meishan, computing power is being transformed into a "new farming tool."

Villagers in Yongfeng Village, Dongpo District, Meishan, are making use of the idle time after rice harvest to plant vegetables quickly.

High yields that can be calculated accurately

Zhao Youyong, who has farmed for over a decade, has contracted more than 800 mu of land in Yongfeng Village. Since last year, two new tools have appeared in his fields: a smart rice transplanter capable of autonomous driving and a rice-specific AI large model. "In the past, transplanting rice seedlings required two people. Now the smart transplanter can drive itself, saving at least ten to twenty yuan per mu," Zhao Youyong said. The bigger change is in field inspection. "The AI large model automatically determines pests, diseases, and deficiencies in fertilizer or water. We directly target the issues based on feedback, saving a lot of time."

"Previously, providing technical plans for large growers was very energy-consuming. For just a few dozen mu, we needed a whole plan. With the AI large model, precision is high, and experts only need to manually review, greatly improving efficiency," said Yang Zhiyuan, deputy director of the Sichuan Rice New Variety and New Technology Pilot R&D Platform. He also admitted that the rice-planting AI large model has so-called "hallucinations," sometimes confusing variety types and regional climate differences and giving inappropriate suggestions. "This requires us to differentiate." According to him, the large model is still in the pilot testing stage. Well-educated large growers have high acceptance, while older smallholders still need further training.

Rice harvest in Yongfeng Village, Dongpo District, Meishan.

Behind the high yield is the integrated application of key technologies such as precision strip sowing and seedling raising, "base-slow-release plus fast topdressing" fertilization, and side-deep placement of slow-release fertilizer. "The yield met expectations. Although it did not reach the historical high, production costs dropped significantly, and before harvest we used multispectral imaging combined with models to estimate yield, so we had confidence," Yang Zhiyuan said. "Combining improved seeds and methods with AI precision management can effectively unleash rice yield potential," said Ma Jun, a professor at Sichuan Agricultural University. He stated that AI rice-planting technology will be further refined so that computing power becomes a "new farming tool" for the "Tianfu Granary."

Zhao Youyong firmly believes this: "When I was a child, farming relied entirely on manual transplanting, harvesting, and drying. Now it's all transplanters, harvesters, and dryers. In another 10 or 20 years, artificial intelligence will definitely develop even better. That is inevitable."

Good medicine coded into existence

Starting from Yongfeng Village, the wave of digital transformation flows to Han'an Village, Xieja Street, Pengshan District. The local "rice + Alisma" rice-medicine rotation has continued for decades but has long been trapped by the problem of "growing it but not selling it at a good price."

Pengshan Alisma GAP Core Demonstration Zone.

The change began with a "code." Today, more than 3,000 mu of Alisma in Han'an Village are fully incorporated into GAP (Good Agricultural Practice for Chinese Medicinal Materials) closed-loop management, and each batch of medicinal materials has a unique "digital identity card." Scanning the code on a mobile phone reveals field growth images, fertilization and pesticide records, and quality inspection reports at a glance. It is clear and traceable which field the medicinal materials came from and what processes they went through.

"In the future, products without standardized production and traceability will be hard to sell," said Zhang Yonghong, Party Secretary of Han'an Village. He noted that the state's requirements for traceability of Chinese medicinal materials are becoming stricter. With a digital identity card, pharmaceutical companies have higher recognition when purchasing, "and they can sell at a better price."

Large grower Luo Chaoqing feels this keenly. He has planted 500 mu in Han'an Village. Compared with rice alone, rice-medicine rotation increases income by at least 1,000 yuan per mu. "The digital identity card, on one hand, enhances product visibility. On the other hand, the quality of medicinal materials can meet national requirements, so the price will definitely be higher," Luo Chaoqing said. "I can work the land and also make money."

From "growing well and being traceable" to "selling well," there is still one processing step missing. "In the past, after harvesting Alisma, it had to be transported to Xichang for drying. A truck of more than 30 tons incurred over 4,000 yuan in freight costs," Luo Chaoqing explained. The village's fresh-processing center is about to be put into use by the end of the year, integrating washing, drying, and cold storage, and directly connecting with slice factories, greatly saving freight costs. Li Hang, deputy general manager of Pengshan Zhengxing Agricultural Development Investment Co., Ltd., the operator of the processing center, said that after production begins, the center will effectively extend the industrial chain, reduce costs for farmers by 3% to 5%, and thus increase returns.

Good farmland that can be seen

The digital revolution above the fields is not limited to the farmland itself; it also extends into vast space. In 2023, Meishan-based enterprise Huan Tian Wisdom launched a self-developed project to build the "Star-Surveyed Tianfu Granary" platform using its own satellites. In 2025, after further iteration of technology and content, the Sichuan Provincial High-Standard Farmland Construction "One Map Manages Farmland" platform officially began pilot promotion, with provincial coverage expected by the end of the year.

At present, this integrated "sky-ground" monitoring network has covered 2.3263 million mu of cultivated land, 1.84 million mu of permanent basic farmland, and 1.6 million mu of high-standard farmland across the city. Relying on BeiDou, remote sensing, and AI algorithms, the system automatically identifies risks of non-grain cultivation of farmland, abandonment, and illegal occupation, and monitors crop growth, meteorological and soil moisture conditions, and soil fertility. The derived "One Map Manages Farmland" system connects the entire chain of digital supervision for high-standard farmland planning, construction, acceptance, and maintenance, filling the gap in full-process digital supervision of high-standard farmland in the province.

"This is a supervision system that achieves 'sky-ground-human' coordination and covers the entire process of high-standard farmland," said He Jiabei, product manager of the Agricultural Business Division at Huan Tian Wisdom. The platform integrates satellite remote sensing, drones, BeiDou positioning, and AI algorithms, covering the full lifecycle of planning, construction, acceptance, and maintenance. In the planning stage, AI compares design drawings with satellite imagery to identify risks of non-compliant site selection. "In the planning stage, survey costs range from 200,000 to 500,000 yuan; in the construction stage, for projects worth 100 million yuan, false reporting losses of 3 to 8 million yuan can be avoided, and annual inspection costs of 100,000 to 200,000 yuan can be saved," He Jiabei said. In the construction stage, satellites record construction progress, BeiDou work badges record personnel trajectories, and AI cameras monitor the duration of machinery operations. In the acceptance stage, satellite imagery is compared with as-built drawings to identify risks such as substandard engineering quantities.

"Star-Surveyed Tianfu Granary" platform.

In the maintenance stage, the platform focuses on monitoring risks such as "non-grain" and "non-agricultural" use and abandonment. He Jiabei gave a real case: in a high-standard farmland project in Guiping Town, Meishan area of Sichuan Tianfu New Area, satellites identified that cultivated land had been converted into an orchard. After the system pushed the alert, the local agriculture bureau assigned it to the village committee and maintenance unit. After on-site verification and rectification, grain was planted again. "Previously, manual field patrols were used. Sichuan has many mountains, and many places could not be reached. Now satellites photograph from the sky and field chiefs inspect on the ground. Problems are pushed with one click, forming a closed loop," said Wang Sen, an operations and maintenance engineer at Huan Tian Wisdom. What impressed him most was a monitoring event in October 2023: near Shigao Street, rice growth was poor, and "the closer to the city edge, the worse; the farther away, the better." After the alert was pushed, the agriculture bureau, together with environmental protection and housing and construction departments, investigated and found it was due to water and soil impacts brought by urban construction. After timely intervention, conditions improved.

In September's Meishan, rice waves surge. In Zhao Youyong's rice fields in Yongfeng Village, the wheel tracks of the smart transplanter are still clearly visible. In Luo Chaoqing's rotation fields in Han'an Village, Alisma seedlings are being transplanted. On the dispatch screen of Huan Tian Wisdom, the plots of 2.3263 million mu of cultivated land jump in real time. From "relying on the weather" to "data-driven farming," from experience to data, from manual field patrols to satellite protection of fields, the millennia-old Tianfu Granary is being revitalized with the empowerment of digital technology.