In July 2021, record-breaking rainfall inundated Henan Province in central China, causing devastating floods that overwhelmed urban infrastructure, claimed nearly 400 lives (most in the provincial capital of Zhengzhou), displaced close to 14 million residents, and caused an estimated 16.5 billion USD in direct economic losses. The 2021 Henan flood was among the deadliest in recent Chinese history and represented a stark warning about the accelerating risks of climate-related disasters.
As the catastrophe unfolded, what struck me—beyond the sheer scale of the devastation—was the way ordinary citizens were responding. Beyond the headlines and state pronouncements, people were actively voicing their fears, frustrations, and expectations. They weren’t protesting in the streets or writing op-eds; instead, they turned to a state-run petition platform known as the Local Leaders’ Message Board (LLMB)—a government-operated site where citizens submit public queries and receive official responses.
While most of my body of works prior to this study focused on how the incentives and constraints of politicians and bureaucrats shape behaviors and outcomes in a top-down fashion, I began to ask: Could the Henan flood have catalyzed a broader shift in citizen demand for state-led climate adaptation? And if so, how were these demands articulated in a country where formal channels for political participation are tightly controlled?
These questions inspired the research now published in Communications Earth & Environment. My study uses a mixed-methods approach to explore how citizens in China engaged politically through the LLMB to express adaptation-related concerns—even without climate framing or direct exposure—in the wake of the 2021 Henan flood.
Climate adaptation in constrained contexts
Scholars and practitioners alike have increasingly recognized that climate adaptation—measures taken to reduce vulnerability to climate impacts—is just as crucial as mitigation. But research and practice have historically lagged in this area, especially in the Global South and in authoritarian states. In such settings, limited resources and restricted civil liberties often hinder both policy development and public participation. Yet unlike mitigation—which focuses on reducing emissions at large scales—adaptation requires context-specific knowledge, making public participation especially vital for identifying localized risks and priorities.
Most existing literature on climate opinion and adaptation behavior comes from democracies, where voting, protest, and advocacy provide visible channels for influence. In autocracies, citizen preferences are harder to trace and often presumed to matter less—and not without reason. Yet China’s governance system includes mechanisms for what some call “responsive authoritarianism”—state-sanctioned platforms that absorb grievances without challenging political authority.
LLMB is one such platform. Operated by People.cn, the online arm of the Chinese Communist Party’s flagship newspaper, the LLMB allows citizens to submit petitions to local government leaders. While messages are moderated and subject to censorship, the platform is far from symbolic. Many local officials actively respond, and research suggests LLMB can function as both a bureaucratic accountability tool and a release valve for public frustration. This makes LLMB an analytically rich—if imperfect—window into how citizens express policy demands under authoritarian rule.
Tracking adaptation concerns through petitions
Drawing on a large dataset of LLMB petitions submitted nationwide during 2021, I traced how public concern shifted in the wake of the Henan flood. I defined climate adaptation-related petitions as those referencing flood risks, drainage issues, or infrastructural vulnerabilities—whether for immediate protection or longer-term resilience.
My analysis focused on whether the volume and content of these adaptation petitions changed after the Henan flood. Using a dynamic difference-in-differences design, I found a sharp and sustained increase in adaptation-related requests in Henan immediately after the flood. Crucially, this shift was not short-lived emotional venting: elevated levels persisted for weeks, indicating a lasting change in public concern.
Moreover, adaptation petitions also rose in other provinces that were not directly affected by the Henan flood. Citizens in cities like Kunming, Chengdu, and Shenzhen explicitly referenced the disaster when raising concerns about their own drainage systems, embankments, or urban flood preparedness. This suggests that major disasters can produce "spillover" effects, mobilizing adaptation demands even in distant regions through media coverage and shared risk awareness.
A pragmatic grammar of resilience
What did these adaptation demands look like in practice? To answer this, I employed neural topic modeling (BERTopic) on the Henan petitions. The resulting themes were mostly grounded in local, tangible concerns: waterlogged basements, crumbling culverts, unsafe bridges, and overwhelmed sewers. Citizens rarely used abstract climate terms like “global warming” or “climate change.” Instead, they expressed what I call a “pragmatic grammar of resilience”—a language focused on specific safety hazards and infrastructure needs.
This raises a point counter to some existing scholarship: climate adaptation doesn’t always require “climate awareness” in the conventional sense. People don’t need to invoke scientific concepts to demand climate-relevant action. In authoritarian systems, such grounded, depoliticized framings may even be more effective in prompting state attention than ideologically loaded terms.
From voice to governance
Of course, expressing demands is not the same as achieving policy change. My study focuses on citizen voice, not government response (which I reserve for future analysis). But the act of petitioning—especially in the aftermath of a disaster—can shape bureaucratic priorities and surface blind spots in top-down planning.
In Zhengzhou, for instance, over $8 billion USD had been invested in “sponge city” infrastructure—a national pilot initiative designed to improve urban flood resilience. Yet the July 2021 flood revealed that many local risks, such as inadequate metro station drainage or neglected embankments, had gone unaddressed. Petitioners flagged these gaps with precision. Their concerns reflected place-specific vulnerabilities that standardized, top-down, technocratic designs had overlooked.
This points to a powerful insight: citizen feedback, even when institutionally constrained, can improve the relevance and cost-effectiveness of adaptation policy. In rapidly urbanizing autocracies like China, adaptation should not be viewed solely as a top-down technical exercise. Rather, it is a shared process of governing risk, in which citizens—through formal or informal means—play an essential role.
This project was intellectually challenging. It required navigating complex methodological terrain, self-training in causal inference and machine learning-based topic modeling, working with large-scale administrative data, and interpreting political behavior within a context that resists conventional democratic practices.
Yet it also left me with a sense of cautious optimism. Even within tightly controlled systems, climate disasters can galvanize public concern and elicit meaningful, if indirect, engagement. People’s everyday experiences—of flooded homes, stalled metros, or rising riverbanks—can powerfully motivate demands for safety and change.
Moving forward, I hope this research encourages further exploration of climate governance in authoritarian settings. Questions remain. For example: How do governments respond to these petitions? Do public demands translate into policy shifts or budgetary commitments? How do citizen expressions vary across platforms—from official petitions to commercial social media?
Answering these questions will be vital to understanding not only the politics of climate adaptation in China, but also the broader challenge of building resilient societies under authoritarianism—a governance style that had historically prioritized top-down technical solutions. This is especially critical for a policy area like climate adaptation, which depends on bottom-up inputs and iterative, community-informed learning.