From July 4 to 5, Luding County in Ganzi Tibetan Autonomous Prefecture, Sichuan Province, China, experienced the most intense rainfall event of the year. Sudden debris flows struck 12 ravines across six townships, including Lengqi Town, resulting in damage to roads, houses, and farmland. Thanks to precise meteorological early warnings, efficient government coordination, and interdepartmental collaboration, all 1,165 residents from 393 households were evacuated with zero casualties.

Luding County is characterized by a rugged landscape of towering mountains and intersecting gorges. The vertical elevation between the deeply incised Dadu River Valley and the summit of Mount Gongga reaches 6,576 meters, posing three world-class challenges for meteorological monitoring. First, terrain obstruction distorts radar echoes, and traditional monitoring equipment is prone to "blind spots" in the valleys. Second, Localized severe convective weather can develop swiftly and move along complicated paths, leaving extremely limited response time for coupled events of short-duration heavy rainfall and debris flows. Third, pronounced mountain microclimatic variations mean that rainfall intensity can differ greatly between towns under the same weather system, making it difficult to define precise early warning thresholds.

During this severe weather event, Luding Meteorological Service utilized mid- and short-term forecasting models such as “CMA-GFS V_4.0”  “SWC-WINGS” and “Fengqing” to predict the heavy rainfall trend three days in advance and issue a rainstorm warning 24 hours prior, gaining critical time for town-level evacuation organizing. By deploying a network of X-band dual-polarization radars, the Service overcame terrain obstruction limitations, enabling minute-scale detection of precipitation particle types, vertical airflow, and wind field structures. Combined with the “Fenglei” nowcasting model, which integrates wind profiler radar, ground-based observations, satellite imagery, and other multi-source data, the Service was able to accurately capture the development of intense rainfall in the disaster area. In the nowcasting phase, the “Fenglei” model’s echo extrapolation locked onto core impact areas such as Lengqi and Xinglong towns 49 minutes in advance, prompting the issuance of an Orange Rainstorm Warning. Based on minute-scale rainfall data and real-time model outputs, the Service decisively upgraded the warning to Red 20 minutes ahead of the event. Early warning information was disseminated directly to grassroots levels via phone-based call-and-confirm, SMS platforms, and coordination with the Flood Control Headquarters. Upon receiving the alert, Lengqi Town immediately mobilized 54 officials to urgently evacuate 550 residents from 170 at-risk households. All individuals were safely relocated before the debris flow struck.

This case has been recognized by the Ministry of Emergency Management of the People’s Republic of China as one of the typical successful risk avoidance cases in 2025 and promoted nationwide. By leveraging forecasting models such as “Fengqing” and “Fenglei” and adopting a ‘tiered alert plus grassroots call-and-confirm’ approach, this operation stands as a representative example of China’s disaster early warning system. It highlights the critical role of grassroots meteorological services in disaster prevention and mitigation during extreme weather events.