Inundation mapping using hydraulic modeling with high-resolution remote sensed data: a case study in the Acre River Basin, Brazil

• Published in: Modeling earth systems and environment 2024-06, Vol.10 (3), p.3051-3066
• Main Authors: Antunes da Silva, Larissa, Rudorff, Conrado, Ovando, Alex, Pimentel, Alan, Cuartas, Luz Adriana, dos Santos Alvalá, Regina Célia
• Format: Article
• Language: English
• Subjects: Accuracy; Aerial surveying; Aerial surveys; Chemistry and Earth Sciences; Climate change; Computer Science; Earth and Environmental Science; Earth Sciences; Earth System Sciences; Ecosystems; Environment; Environmental impact; Environmental risk; Flooding; Floodplains; Floods; High resolution; Historic floods; Lidar; Mapping; Math. Appl. in Environmental Science; Mathematical Applications in the Physical Sciences; Modelling; One dimensional models; Physics; Review Article; Risk management; Risk reduction; River basins; Rivers; Root-mean-square errors; Roughness coefficient; Satellite imagery; Simulation; Statistics for Engineering; Steady state; Surveys; Terrain models; Urbanization; Water depth
• ISSN: 2363-6203; 2363-6211
• DOI: 10.1007/s40808-024-01972-x

Considering the impacts of climate change in recent decades that have exacerbated the frequency and intensity of floods worldwide and especially in the Amazon region. The city of Rio Branco, located in the southwest of Brazil's Amazon region, has been severely affected by a combination of urbanization in flood-prone areas and increasing floods in the Acre River. In addressing this challenge, accurate inundation mapping plays a pivotal role in shaping effective flood risk reduction strategies. This study employs hydraulic simulations of the Acre River, utilizing the HEC-RAS 1D model. The modeling process integrates a high-resolution digital terrain model, acquired through Light Detection and Ranging (LiDAR), and a rich dataset encompassing conventional and unconventional information from the three most significant historical floods in Rio Branco in 2012, 2015, and 2023. To ensure the precision of the simulations, calibration of the roughness coefficient was conducted for steady-state scenarios, drawing upon a diverse range of observed stream-gauge data. The evaluation of water elevation in steady state reveals an impressive mean error of just 0.01 m, underscoring the model's accuracy. Moving beyond steady-state simulations, the study evaluates unsteady-state scenarios by calculating the root mean square error (RMSE). Results showcase commendable accuracies of 0.22, 0.25, and 0.26 m for the 2012, 2015, and 2023 floods, respectively. Flooding extent simulation was assessed using the Critical Success Index (C) from two optical aerial survey images recorded during the 2012 and 2015 floods and one optical satellite image recorded during the 2023 flood. The accuracy was 0.88 (2012) and 0.84 (2015) for optical aerial survey images and 0.97 for the optical satellite image (2023). Images from Google Street View recorded after 2012 flood event containing high-water marks were used to evaluate the accuracy of maximum water depth simulation along the floodplain and presented a mean error of 0.17 m.