The story behind this project starts with Dr. Peirong Lin’s observation and insights into the urban development in levee-protected floodplains when trying to understand the human perturbations to the river-floodplain system (Fig 1). Levee construction often provides a sense of security that may promote urban development and agricultural activities in its protected floodplains. At the same time, it can also reduce people’s awareness and preparedness of potential flood risks, which may end up increasing socioeconomic losses from exposures to high-consequence low-probability flood events. Echoing this logic, existing studies have speculated that levee construction might be one of the promoting factors for the increased population exposed to floods over the years. However, an objective quantitative assessment of the interplay between levee construction and floodplain urban expansion is still lacking.

Fig 1. The satellite images of four selected sites hinting urban development within the floodplain (from left to right: Larksville, PA; Louisville, KY; Chico, CA; Albuquerque, NM).

Since October 2021, Meng Ding from the Department of Geography and Geospatial Sciences at Kansas State University visited Dr. Lin as a visiting student at the School of Earth and Space Sciences, Peking University. Since then, we started to gather data and build methods aiming to gain an in-depth understanding of this interplay.

Through earlier experimentation, we recognized the complexity of factors that collectively contribute to urban expansion, which made it challenging to disentangle the levee effect (Fig 2). Also, data constraints and the scattered distribution of levees could undermine a clear macro-scale picture of the levee effect.

Fig 2. Raw urban expansion curves of selected counties. a, b, and c are counties with significantly faster urban growth after levee construction. d, e, and f are opposite examples that the urban growth rate becoming smaller after the levee was built. g, h, and i show that urban areas did not smoothly grow in some counties, which suggests data limitations and more complex urban growth patterns.

Therefore, to better disentangle the association between levee construction and people’s perceptional change in flood risks, we introduced the composite analysis and similar concepts of econometric methods using thousands of levee systems from the National Levee Database (NLD) and ~70 years of annual land-use land-cover change (LULCC) data from the USGS, which helps to reveal the rate of urban expansion in the floodplain protected by levees has accelerated more than its surrounding areas (Fig 3), indicating a clear perceptional shift in treating floodplains as more habitable land than other regions after levee construction. We went on to assess the spatio-temporal patterns of the levee effect by breaking down the analysis into different watersheds and states, showing that the central and western US can exhibit a stronger levee effect. Analysis of the temporal dynamics suggests that effective regulatory and legislative measures in the US may have gradually reversed the levee effect for more sustainable floodplain management (see Figs. 2& 3 in the article for more details).

Fig 3. Figures show the composite analysis of the urban area (10³ km²) in the levee-protected floodplains (U_p) and that in the counties (U_c), respectively. X-axes separate the urban expansion time series into years before and after the levee construction year T₀; K shows the linear urban expansion rate.

Through a more comprehensive literature review, we elaborated on the role of levees in flood risk management (see Fig 4 and more details in the article) and discussed the need to conduct more research along the line to promote a rethinking of human-water relationships in the future.

Fig 4. Conceptual diagram illustrating the role of levees in flood risk management. ①②③ refers to the links for levee effect to affect three determinants of flood risks (i.e., hazard, exposure, vulnerability). “+” and “–” denote positive/negative correlation, respectively.

To the best of our knowledge, this study is one of the first attempts to move the conceptualization of the levee effect to quantitative assessments, where interesting spatio-temporal patterns are revealed. Our study also promotes the understanding of how the combined effects of “soft path solutions” and structural measures can influence sustainable floodplain management. Although the levee effect defined in our study as the floodplain urban expansion associated with levees may be different from the more desirable definition that eventually involves the associated risks, we anticipate this study to open up discussions for more research on the complex interplay between hydraulic engineering infrastructures and floodplain development. Through addressing difficulties when conducting this work, we also realize the need to advocate for the development or open sharing of levee databases, the risk database, as well as the development of diversified adaptation strategies for different countries based on their development stage and hydroclimatic contexts.

The contributing authors of this work also include Drs. Shang Gao, Jida Wang, Zhenzhong Zeng, Kaihao Zheng, Xudong Zhou, Dai Yamazaki, Yige Gao, and Yu Liu. This study is supported by the Open Research Program of the International Research Center of Big Data for Sustainable Development Goals (Grant No. CBAS2022ORP05) and the Fundamental Research Funds for the Central Universities, Peking University on “Numerical modeling and remote sensing of global river discharge” (No. 7100604136). For more details of this work, please see our article entitled Reversal of the levee effect towards sustainable floodplain management at: https://doi.org/10.1038/s41893-023-01202-9