Flood protection in a changing society: a perspective from historical agriculture transformation during 1931 and 1954 floods in Yangtze River Basin, China

As the longest river in Asia, the Yangtze River has shown its impact on human societies with floods recorded since 12th century. In 1931, the Yangtze River has manifested its force again with one of the deadliest floods ever recorded in Chinese history, causing 422,499 deaths, damages to more than 25.2 million people and 58.7 billion m2 farmland. Similar flood occurred again in 1954, resulting in 31,762 deaths, damages to 18.9 million people and 31.7 billion m2 farmland. Researches have shown that 1954 flood being larger and higher compared to 1931 flood. However, it is still unclear for what reason that a more severe flood leading to less damage. Was it because of the change of policy and what can we learn from it? To answer this question, first, we constructed a conceptual framework of 1930s and 1950s agricultural society. Then, the flood inundation process of 1931 and 1954 floods was reconstructed with gauged rainfall dataset. After referring to the investigation report, the inundation information was applied to re-estimate the flood damage on farmland, houses, and residents under different scenarios. Furthermore, apart from traditional countermeasures, the effect of society transformation with floods were estimated using the potential crop production (PCP). Finally, an agent-based model (Farmer Landlord Inundation Production, FLIP) was constructed to simulate the agricultural transformation and its impact on residents’ response. The results have shown that traditional countermeasures were of certain effect. For example, the reinforcement of levees in 1950s was more effective in reducing inundation area of 8% compared to 1931, while the construction of detention basins accounted for 2%. However, with only traditional countermeasures failed to explain the relative success of agricultural product in 1954. Which, according to PCP result, the observed rice production was 12% higher than potential in 1954, while it was 29% lower in 1931. Here we assumed such difference could be ascribed to drastic society transformation in 1930s and 1950s (e.g., increase of absentee landlords in 1930s, land reform movement in 1950s). The effect of which was partially demonstrated with FLIP model, indicating an increase of crop production after eliminating landlords during floods. Our results demonstrate how society transformation are likely to affect the damage of and response to floods in a different (sometimes more important) way from traditional countermeasures in modern Chinese history. We anticipate our research to be a starting point towards deeper understanding of human and hazard, and the knowledge of which is likely to be applicable to many other regions and times.