Interview: Extreme Disasters Are Upon Us, Yet We’re Still Learning the Hard Way

Rivers in the South’s “humid zone”, which carry a higher baseline flood risk, generally do not exhibit such extreme variability. During the flood season, flows may reach several thousand or even tens of thousands of cubic metres per second, while outside this period they still maintain 2,000–3,000. The floodplains might be inundated once every few years. Yet extraordinary floods with return periods exceeding 20–40 years are far harder to guard against, as practical flood-fighting experience is rarely passed down through generations.

The rise in extreme rainfall and exceedance floods is also transforming how government bodies operate. “Call-and-confirm” alerts have become the primary tool for emergency management departments dealing with torrential rain and flooding. Whenever a red warning is issued or a flood response is activated, officials must make sequential phone calls down to the grassroots level: “Heavy rain is imminent. Evacuate residents in high-risk areas immediately!” For those navigating this protocol for the first time, initial chaos is almost inevitable.

As climate change pushes the 400 mm isohyet northwards, the probability of extreme rainfall triggering major and catastrophic flash floods, as well as flooding in small and medium-sized rivers, will continue to rise in the mountainous valleys and gully terrains located between China’s 400 mm and 800 mm isohyets.

In August 2024, widespread torrential rain and flooding struck Jianchang County and other areas under Huludao City, resulting in at least 11 fatalities. A rescue team leader who took part in both last year’s floods in Huludao, Liaoning, and this year’s flooding along the Beijing–Chengde border told me that, in terms of type, scale, and on-the-ground conditions, the two disasters bore a striking resemblance.

Can we draw lessons from the disaster-stricken regions mentioned above? Can we recognise that it could happen to anyone, and then give serious thought to how we should respond?

Hao Nan: In 2013, drawing on earlier experiences, I put forward the concept of the ‘five major imbalances’: in how different disaster types are perceived, in the distinction between major and minor events, between urban and rural areas, across the various phases of recovery, and between pre- and post-disaster periods. These five imbalances in public attention persist today with little fundamental change; we have simply shifted from focusing primarily on earthquakes to paying more heed to meteorological disasters. Resource allocation has always tracked public interest far more closely than the actual scale of the damage. And rural areas have consistently received less attention than their urban counterparts.

Over the past decade and a half, the trend of climate change driving more frequent extreme weather has been unmistakable. Yet it took the devastating floods in Henan in 2021—officially logged as the ‘23·7’ storm, though the media largely branded it the ‘720’ downpour—to truly wake the public to the reality. In the Zhengzhou city centre, 200mm of rain fell within a single hour on the evening of 20 July. The unprecedented deluge naturally triggered severe urban waterlogging, and news that more than a dozen commuters had perished in the subway tunnels ignited immediate public outrage. Far fewer people, however, were aware of the even more severe casualties that occurred around the same time—or indeed, earlier—in the surrounding regions.

The officially recorded death toll from the Zhengzhou floods stands at 380. Seventy per cent of those fatalities occurred in the hilly counties and municipalities on the urban fringes, with most deaths taking place before 13:00 on 20 July. The vast majority lost their lives to flash floods, geological hazards, and river flooding in smaller waterways; only around one in ten perished in urban waterlogging. Yet public discourse has focused almost exclusively on the latter.

On the far side of the Yellow River, Xinxiang experienced extreme rainfall in the days surrounding 20 July that was no less severe than Zhengzhou’s. The death toll was lower—largely because the terrain is predominantly flat, allowing floodwaters to move more slowly and giving people time to evacuate or be rescued. But the area was effectively turned into a vast inland sea. Satellite imagery revealed vast stretches submerged; villages were reduced to isolated islands. Our initial estimates suggest that, including downstream areas in Qixian and Junxian counties of Hebi City, nearly one million people were trapped by floodwaters, with rescue teams reaching at least 400,000 of them. In Weihui alone—a county-level city under Xinxiang’s administration—200,000 residents had to be completely evacuated.

Hao Nan: All these years on, certain crucial points remain largely unaddressed. The mechanisms and risk profiles of urban waterlogging are fundamentally different from those of flash floods, geological hazards, and river flooding. The risk of fatality from urban waterlogging is significantly lower than that posed by flash floods and river surges, and the long-term impact of the latter is invariably more profound. Even in Zhengzhou, where rainfall was so extreme, the urban standing water cleared within a day. Meanwhile, towns and villages along the downstream rivers remained submerged for over half a month.

Furthermore, the water accumulating in cities is predominantly clean rainwater, making cleanup relatively straightforward. Urban infrastructure typically bounces back within days, and residents rarely suffer a permanent loss of livelihood. By contrast, torrential floodwaters sweeping down from the hills carry vast amounts of silt and debris, leaving behind total destruction of homes and livelihoods in their wake.

During the catastrophic floods in Qiandongnan, Guizhou, this past June, the floodwaters in Rongjiang City receded to leave behind up to two metres of silt mixed with debris. Crucially, mountain torrents also obliterate farmland, roads, and vital infrastructure. In the hard-hit townships of Xinglong County, Hebei, which we recently visited, extensive tracts of prime arable land in mountain valleys, floodplain fields, and terraced orchards on hillsides were completely washed away. When prime farmland is buried by mudflows, it is extremely difficult to restore and must be officially reassessed. Although affected farmers receive some financial compensation for lost land, the damage to their means of livelihood is effectively irreversible.

This trend of increasingly frequent mountain flooding will only persist—and likely accelerate—in the years ahead.

II. Disaster Mitigation Outreach: Who Can Truly Afford to Spend Three Months Drinking with the Villagers?

‘Disaster resilience’ and disaster prevention and mitigation are not conceptually identical, but that does not preclude them from being addressed in tandem during practical work. At the very least, it allows for a functional dialogue. As for charitable donations towards disaster preventionmitigation, funding has not increased over the past decade; it remains scarce, with very few active projects.

Measuring the return on investment for prevention and mitigation is notoriously difficult, largely because there is no scientifically robust benchmark. Moreover, the more effective these measures are, the closer post-event conditions resemble business as usual, making the impact inherently invisible. Without tangible returns, investment dries up. The media and public attention naturally gravitate towards actualised losses, whereas averted losses are difficult to quantify and rarely receive serious consideration.

Extraordinary disasters are, after all, low-probability events. If prevention and mitigation efforts are rolled out across a thousand low-risk communities, there is a strong likelihood that no severe disasters will materialise within the operational lifespan of those measures. To enhance the efficacy of such investment, we must first accurately identify high-risk communities—those with a statistically higher probability of suffering severe losses.

Disaster prevention and mitigation typically command serious attention only in the aftermath of a catastrophe. Yet, it is exceedingly difficult to instruct communities on how to manage crises they have never witnessed. The very events that inflict the most devastating losses are often historic-scale disasters that lie entirely outside local lived experience.

In most circumstances, lessons drawn from past disasters are invaluable. However, when confronted with extraordinary events, this reliance on historical precedent can actually become a hindrance. This year, in the flood-affected regions of Meizhou and Longyan, we observed how residents living along riverbanks traditionally responded to rising waters by relocating valuables and elderly family members with limited mobility from the ground floor to the first floor. But during last June’s exceptional deluge, the water levels crested well above the first-floor windows…

Only a handful of organisations with long-term rural commitments can fold disaster prevention and reduction into their existing work. You could count them on the fingers of one hand, and they aren’t necessarily experts on disasters themselves. When some organisations are just getting ready to launch community disaster reduction programmes, they’ll ask, “Do you have a presentation? Is there a ready-made curriculum?” But it’s not an off-the-shelf package.

Take evacuation, for instance. The aim is simply to get people away from danger. But the hazard sources in a community—say, landslides, rockfalls, or torrential rain upstream in a river basin—come with specific characteristics and dynamic mechanisms. Figuring out their nature, how they propagate, and what area they’ll hit next demands highly specialised technical expertise. All of this dictates when to evacuate, which routes to take, and how to draft contingency plans. Different hazards require entirely different approaches, and every community’s baseline conditions vary wildly. Add in the myriad factors that shape what residents know, trust, and ultimately act upon, and the complexity multiplies.

Lately, many organisations have moved into community climate adaptation, usually starting with an assessment of meteorological disaster risks. Yet the general feedback is that this part is notoriously difficult to execute. From what I’ve seen, it’s incredibly hard to actually bake risk reduction or resilience-building into the tangible deliverables of a project.

Governments at every level now place a heavy premium on safeguarding lives. The groundwork for meteorological disaster response is tightly enforced, with accountability cascading down through every tier. Departments covering meteorology, water resources, emergency management, and natural resources are all pressing local authorities to rigorously implement disaster prevention and evacuation measures. For instance, under the Ministry of Natural Resources’ directives, communities are required to implement a dual-control approach for geological hazards—managing both identified danger sites and broader risk zones—alongside issuing quick-reference guidance cards for disaster avoidance at specific hazard points. In theory, rural communities are also meant to establish publicly funded community roles such as disaster information liaisons and geological hazard monitors.

Hao Nan: But putting it into practice is just as difficult. During torrential rain and flooding, floods and debris flows in the source areas often arrive simultaneously with the early warnings, which stands in stark contrast to what most emergency response plans state. Whether a community disaster prevention and mitigation project can develop practical solutions during peacetime that genuinely resolve issues when a disaster strikes marks the true dividing line of its professionalism. At times, community initiatives may succeed in raising awareness, but their actual effectiveness in preventing or mitigating disasters remains questionable.

On one occasion, an organisation invited us to conduct a disaster risk assessment for a village. Given the complex risk profile there, it required bringing in three or four specialists from relevant disciplines to carry out on-site surveying and evaluation. This involved pinpointing the toe and crown of a landslide mass — noting that fissures and uplifts serve as precursors — assessing the geotechnical composition, and determining the thickness and volume of the potentially mobile soil layer, its likely direction of movement, and which households would be affected. The budget for an assessment like this would rarely exceed 5,000 yuan.

Although the funding was limited, we were more than willing to take part in such a meaningful project, yet we have not encountered another since.

Another category comprises community disaster capacity-building initiatives, which are typically bundled as supplementary components of post-disaster relief and are usually one-off. A friend recently travelled to Dingri County in Tibet to carry out a community seismic resilience project; funding had been allocated following a 6.8-magnitude earthquake earlier in the year. Such projects follow a consistent pattern: they are only implemented in locations that have already been struck, and typically within a single year. They serve mainly to offer survivors a sense of security and psychological solace. After two years, even the locals lose the will to continue.

Prevention and mitigation projects, which rightly belong in the pre-disaster phase, are relegated to the aftermath. Relying on actual disasters to identify high-risk communities is simply too costly.

III. Boulders Tumbling Downhill: Louder Than Village Officials Shouting