FLOOD SUSCEPTIBILITY IN BOANE (MOZAMBIQUE): A RANDOM FOREST AND GEOTECHNOLOGICAL APPROACH
DOI:
https://doi.org/10.56238/revgeov17n5-084Keywords:
Floods, Susceptibility, Random Forest, GeotechnologiesAbstract
This study aimed to model and map flood susceptibility in the District of Boane (Mozambique), considering the increasing occurrence of extreme events associated with climate change. For this purpose, the Random Forest (RF) machine learning algorithm was applied, integrating 12 conditioning variables of climatic, topographic, hydrological, anthropogenic, and spectral nature. The flood inventory was developed using Landsat 9 imagery with the support of the Modified Normalized Difference Water Index (MNDWI), while model validation was performed using Sentinel-2 imagery. The modeling process was carried out in the R environment, with data split into 70% for training and 30% for validation. The results indicated high model performance, with overall accuracy values of 0.93 for the model and 0.98 for the susceptibility map. The most relevant variables were proximity to water bodies and altitude, highlighting the strong influence of geomorphological factors on flood dynamics. The final map revealed a predominance of areas classified as having low susceptibility (59.35%), whereas areas with higher susceptibility were mainly concentrated in the central region of the District of Boane, coinciding with agricultural zones and human settlements. Therefore, it can be concluded that the Random Forest model demonstrated high predictive capability for flood susceptibility mapping, constituting a robust tool for territorial planning and environmental management, particularly in identifying areas potentially exposed to infrastructure damage, agricultural losses, and socioeconomic impacts associated with flooding.study aimed to model and map flood susceptibility in the District of Boane (Mozambique), considering the increasing occurrence of extreme events associated with climate change. For this purpose, the Random Forest (RF) machine learning algorithm was applied, integrating 12 conditioning variables of climatic, topographic, hydrological, anthropogenic, and spectral nature. The flood inventory was developed using Landsat 9 imagery with the support of the Modified Normalized Difference Water Index (MNDWI), while model validation was performed using Sentinel-2 imagery. The modeling process was carried out in the R environment, with data split into 70% for training and 30% for validation. The results indicated high model performance, with overall accuracy values of 0.93 during training and 0.98 during validation. The most relevant variables were proximity to water bodies and altitude, highlighting the strong influence of geomorphological factors on flood dynamics. The final susceptibility map revealed a predominance of areas classified as having low susceptibility (59.35%), whereas areas with higher susceptibility were mainly concentrated in the central region of the District of Boane, coinciding with agricultural zones and human settlements. Therefore, it can be concluded that the Random Forest model demonstrated high predictive capability for flood susceptibility mapping, constituting a robust tool for territorial planning and environmental management, particularly in identifying areas potentially exposed to infrastructure damage, agricultural losses, and socioeconomic impacts associated with flooding.
Downloads
References
Amaral, F., Piroli, E. L., & Santos, V. C. (2024). Mapeamento da suscetibilidade a inundação usando os modelos Floresta Aleatória e Máquina Vetor de Suporte com diferentes tipos de kernel. Geography Department University of Sao Paulo, 44, e213348. https://doi.org/10.11606/eISSN.2236 2878.rdg.2024.213348.
Bicá, H. H. S., Madal, M. A. P., Famba, D. S., & de, J. (2025). Mapeamento das áreas Agrícolas Inundadas na cheia do Ano 2023 usando dados Sentinel-1 e Sentinel-2 no Distrito de Boane.
Breiman, L. (2001a). Random Forests. Machine Learning, 45(1), 5–32. https://doi.org/10.1023/A:1010933404324.
Chicombo, t. F., & Moreira, a. A. M. (2024). Mapeamento de suscetibilidade às inundações com base na análise multicritério (ahp): estudo de caso do distrito de chókwe, em moçambique. Caminhos de geografia, 25(99), 301–319. Https://doi.org/10.14393/rcg259970390.
Choubin, B., Moradi, E., Golshan, M., Adamowski, J., Sajedi-Hosseini, F., & Mosavi, A. (2019). An ensemble prediction of flood susceptibility using multivariate discriminant analysis, classification and regression trees, and support vector machines. Science of The Total Environment, 651, 2087–2096. https://doi.org/10.1016/j.scitotenv.2018.10.064.
Costache, R., Arabameri, A., Elkhrachy, I., Ghorbanzadeh, O., & Pham, Q. B. (2021). Detection of areas prone to flood risk using state-of-the-art machine learning models. Geomatics, Natural Hazards and Risk, 12(1), 1488–1507. https://doi.org/10.1080/19475705.2021.1920480.
Dransfeld, N. M. (2025). Universidade do Estado de Santa Catarina – UDESC - Centro de Ciências Tecnológicas – CCT Programa de Pós-graduação em Engenharia Civil
Islam, M. M., Sado, K., & Sumi, T. (2021). Flood hazard assessment using geospatial techniques and machine learning models. Journal of Hydrology, 592, 125–148. https://doi.org/10.1016/j.jhydrol.2020.125148.
Kaiser et al. (2022). Kaiser, A., Merz, B., & Vorogushyn, S. (2022). A review of flood event data sources for hazard and risk analysis. Natural Hazards and Earth System Sciences, 22(2), 401–425. https://doi.org/10.5194/nhess-22-401-2022.
Luiz Felipe da Silva (2025). Aplicação de Algoritmos de Aprendizado de Máquina para o Mapeamento de Suscetibilidade a Inundações da Bacia Hidrográfica do Rio Itajaí. Universidade Federal De Santa Catarina.
Mhawej, Mario, Ghaleb Faour, Chadi Abdallah, and Jocelyne Adjizian-Gerard. (2016). Towards an establishment of a wildfire risk system in a Mediterranean country. Ecological informatics.
Nhanenge, Edilson. (2024). Modelação da Susceptibilidade à Inundações Baseada Em Análise Multicritério Em Sistemas de Informação Geográfica. Faculdade De Ciências.
Oliveira, W. N. de, Miziara, F., & Ferreira, N. C. (2019). Mapeamento do Uso e Cobertura do Solo de Moçambique Utilizando a Plataforma Google Earth Engine. Anuário do Instituto de Geociências, 42(1), 336–345. https://doi.org/10.11137/2019_1_336_345.
Oliveira, J., Correia, D., verçosa, C., Figueroa, M., Cabral, C., Vitor, L., Francisco dos Santos, A., Martins, A., Teixeira, L., Pereira, G., Anunciação, J., Pereira, F., Vasconcelos, J., Silva, A., Fernandes, N., Silva de Oliveira, P., Kely, N., Santos, R., Barros, J., & Santos, M. (2021). A utilização inadequada do solo e seus impactos na degradação das áreas. Research, Society and Development, 10, e35101219948. https://doi.org/10.33448/rsd-v10i12.19948.
Rahmati, O., Pourghasemi, H. R., & Zeinivand, H. (2016). Flood susceptibility mapping using frequency ratio and weights-of-evidence models in GIS. Geocarto International, 31(10), 1179–1198. https://doi.org/10.1080/10106049.2015.1041559.
Rodriguez-Galiano, V. F., Ghimire, B., Rogan, J., Chica-Olmo, M., & Rigol-Sanchez, J. P. (2012). An assessment of the effectiveness of a random forest classifier for land-cover classification. ISPRS Journal of Photogrammetry and Remote Sensing, 67, 93–104. https://doi.org/10.1016/j.isprsjprs.2011.11.002.
Silva De Oliveira, S., Machado, H. T. D. S., Araújo, K. D. C., Silva, C. D. S., & Ávila, R. W. (2023). Nematodes associated with Leptodactylus cf. mystaceus (Anura: Leptodactylidae) in agricultural landscapes Ibiapaba plateau, Ceará. https://doi.org/10.15446/caldasia.v46n2.101535.
Tamele, J. (2024.). Acesso e alocação de terras para além dos grandes investimentos: o papel das elites políticas e económicas em Boane.
Xie, S., Liu, L., Zhang, X., Yang, J., Chen, X., & Gao, Y. (2019). Automatic Land-Cover Mapping using Landsat Time-Series Data based on Google Earth Engine. Remote Sensing, 11(24). https://doi.org/10.3390/rs11243023.
Yesilnacar, E.K. (2005). The application of computational intelligence to landslide susceptibility mapping in Turkey. Parkville, Victoria: University of Melbourne.
