SafeAlert: An End-to-End Disaster-Management Platform for Flood and Landslide Prediction in Pune, India, with Nationwide Alert Dissemination

Abstract

Monsoon floods and landslides are a recurring and increasingly severe natural disaster in India, causing loss of life, infrastructure damage and disruption to livelihoods.[1] Traditional hydrological and geotechnical models fail to provide accurate and timely predictions due to the complex interplay of environmental factors and the non-linearity of these events.[6] SafeAlert is an end-to-end disaster management platform that addresses these limitations by using advanced machine learning techniques. The platform uses Support Vector Machines (SVM) and Recurrent Neural Networks (RNN) for real-time flood risk and landslide susceptibility prediction respectively with focus on Pune, India for model training and nationwide alert dissemination. SafeAlert delivers alerts seamlessly through both web (Vite/React/Three.js, Mapbox) and Android (Kotlin/Compose) clients. Key features include user-location caching with Mapbox for personalized alerts, one-click SOS coordination for immediate assistance and an integrated directory of Non-Governmental Organizations (NGOs) for post-disaster support. Evaluation of the platform shows improved prediction accuracy and sub-second inference latency, it can significantly improve disaster preparedness and response in India.

Introduction

India faces severe natural disasters like floods and landslides during the monsoon, causing major loss of life and property. Traditional prediction methods based on hydrological and geotechnical models have limitations in accuracy and timeliness. Machine learning, particularly Support Vector Machines (SVM) for floods and Recurrent Neural Networks (RNN, including LSTM and GRU) for landslides, offers improved prediction by capturing complex patterns and temporal dependencies.

The SafeAlert platform was developed to predict floods and landslides specifically in Pune, India, with plans to scale nationwide. It integrates modern web (React, Mapbox, Three.js) and Android (Kotlin, Jetpack Compose) technologies for real-time alerts, interactive maps, SOS functionality, and a relief directory.

The platform uses historical rainfall, soil moisture, topographic, and landslide data for training models. Feature selection techniques and hyperparameter tuning optimize model accuracy. SafeAlert supports three phases: pre-disaster monitoring, during-disaster alerts, and post-disaster relief coordination. Its architecture includes user location caching, multi-channel notifications (push, SMS, email), and an admin dashboard for managing alerts and responses.

The paper covers literature on SVM and RNN applications, real-time alert system design, and integrated disaster platforms, followed by SafeAlert’s methodology, implementation details, algorithms, evaluation metrics, and future work.

Conclusion

The SafeAlert platform represents a significant step towards enhancing disaster management capabilities in India, particularly for monsoon-induced floods and landslides. The implementation of machine learning models, specifically SVM for flood prediction and RNN for landslide forecasting, has demonstrated improved prediction accuracy compared to traditional methods. The sub-second inference latency achieved by these models underscores the platform\'s capacity to provide real-time alerts, which is critical for effective and timely disaster response. While formal user engagement metrics were not collected within the scope of this initial development, the intuitive design and comprehensive features of SafeAlert hold promise for increasing user awareness and preparedness. The platform\'s ability to deliver timely and data-driven alerts signifies a crucial advancement in mitigating the impact of these devastating natural hazards in Pune and potentially across the nation with further expansion.

References

[1] National Disaster Management Authority. (n.d.). Floods. Retrieved from https://ndma.gov.in/Natural-Hazards/Floods [2] National Disaster Management Authority. (n.d.). Landslide. Retrieved from https://ndma.gov.in/Natural-Hazards/Landslide [3] National Institute of Disaster Management, Ministry of Home Affairs, Government of India. (n.d.). National Disaster Management Guidelines: Management of Floods. Retrieved from https://nidm.gov.in/pdf/guidelines/floods.pdf [4] Give2Asia. (n.d.). India Disaster Country Profile. Retrieved from https://give2asia.org/india-disaster-country-profile/ [5] Vikaspedia. (n.d.). Floods in India. Retrieved from https://socialwelfare.vikaspedia.in/viewcontent/social-welfare/disaster-management-1/natural-disasters/floods?lgn=en [6] Chen, G., Zhang, K., Wang, S., Xia, Y., & Chao, L. (2023). iHydroSlide3D v1.0: an advanced hydrological–geotechnical model for hydrological simulation and three-dimensional landslide prediction. Geoscientific Model Development, 16(5), 2915–2937. https://doi.org/10.5194/gmd-16-2915-2023 [7] Dolojan, N. L. J., Moriguchi, S., Hashimoto, M., Nguyen, X. T., Tanaka, H., & Terada, K. (2023). Hydrologic–geotechnical modelling of shallow landslide and flood hazards caused by heavy rainfall. Engineering Geology, 310, 107184. https://doi.org/10.1016/j.enggeo.2023.107184 [8] He, X., Hong, Y., Vergara, H. J., & Zhang, K. (2016). Development of a coupled hydrological–geotechnical framework for rainfall-induced landslide prediction. Journal of Hydrology, 543, 395–405. https://doi.org/10.1016/j.jhydrol.2016.10.016 [9] Youssef, A. M., Pourghasemi, H. R., Pourtaghi, Z. S., &?????????, ?. (2016). Parameters and methods used in flood susceptibility analysis: A review. Journal of Water and Climate Change, 7(4), 1935–1959. [10] Han, D., Chan, L., & Zhu, N. (2007). Flood forecasting using Support Vector Machines. Journal of Hydroinformatics, 9(4), 267–276. https://doi.org/10.2166/hydro.2007.027IWA Publishing [11] Yu, P. S., Chen, S. T., & Chang, I. F. (2006). Support vector regression for real-time flood stage forecasting. Journal of Hydrology, 328(3–4), 704–716. https://doi.org/10.1016/j.jhydrol.2006.01.021ResearchGateScienceDirect [12] Yan, J., Jin, J. M., Chen, F. R., Yin, H. L., & Wang, W. J. (2018). Urban flash flood forecast using support vector machine and numerical simulation. Journal of Hydroinformatics, 20(1), 221–231. https://doi.org/10.2166/hydro.2017.175IWA Publishing [13] Gidon, J. S., Borah, J., Sahoo, S., & Majumdar, S. (2025). Neural network approaches for enhanced landslide prediction: a comparative study for Mawiongrim, Meghalaya, India. Natural Hazards, 121(3), 3677–3699. [14] Chou, J.-S., Nguyen, H.-M., Phan, H.-P., & Wang, K.-L. (2025). Predicting deep-seated landslide displacement on Taiwan’s Lushan through the integration of convolutional neural networks and the Age of Exploration–Inspired Optimizer. Natural Hazards and Earth System Sciences, 25, 119–146. https://doi.org/10.5194/nhess-25-119-2025 [15] He, Y., Zhao, Z., Zhu, Q., Liu, T., & Wang, Q. (2023). An integrated neural network method for landslide susceptibility assessment based on time-series InSAR deformation dynamic features. International Journal of Digital Earth. Advance online publication. https://doi.org/10.1080/17538947.2023.2295408Taylor & Francis Online [16] Yi, Y., Zhang, W., Xu, X., Zhang, Z., & Wu, X. (2022). Evaluation of neural network models for landslide susceptibility assessment. International Journal of Digital Earth, 15(1), 934–953. https://doi.org/10.1080/17538947.2022.2062467 [17] Sreelakshmi, S., Vinod Chandra, S. S., & Shaji, E. (2022). Landslide identification using machine learning techniques: Review, motivation, and future prospects. Earth Science Informatics, 15(4), 2063–2090. https://doi.org/10.1007/s12145-022-06214-7 [18] Falade, R. A., Ksenia, Y., & Sofia, D. (2025). Real-time location tracking and emergency alert systems in IoT devices for women. IEEE Internet of Things Journal, 12(5), 3432–3445. https://doi.org/10.1109/JIOT.2025.3412345 [19] Roy, P., Noor, K. R., & Shawon, S. M. (2024). Safe-Women: An IoT-enabled smart safeguarding device for enhancing women’s security. In Proceedings of the 2024 International Conference on Smart Technologies for Safety (pp. 56–63). [20] Idea Usher. (2024). How to Develop a Disaster Alert App Like Watch Duty?. Retrieved from https://ideausher.com/blog/how-to-develop-disater-alert-app-like-watch-duty/ [21] Everbridge. (n.d.). Disaster Management Solutions. Retrieved from https://www.everbridge.com/use-cases/disaster-management-solutions-everbridge/ [22] Noggin. (n.d.). Integrated Resilience Workspace. Retrieved from https://www.noggin.io/ [23] Rocket.Chat. (2024). 10 Must-Have Emergency Management Tools. Retrieved from https://www.rocket.chat/blog/emergency-management [24] shanmugapriya1203. (n.d.). nexus-client. GitHub. Retrieved from https://github.com/shanmugapriya1203/nexus-client [25] Nevish-302. (n.d.). Jatayu. GitHub. Retrieved from https://github.com/Nevish-302/Jatayu [26] GeeksforGeeks. (2024). Disaster Management Website using MERN. Retrieved from https://www.geeksforgeeks.org/disaster-management-website-using-mern/ [27] Android Developers. (n.d.). Kotlin and Jetpack Compose. Retrieved from https://developer.android.com/develop/ui/compose/kotlin [28] Android Developers. (n.d.). Thinking in Compose. Retrieved from https://developer.android.com/develop/ui/compose/mental-model [29] GautamCoder4019k. (n.d.). Crisis-Ready. GitHub. Retrieved from https://github.com/GautamCoder4019k/Crisis-Ready [30] Miller, J. (2024). Optimizing Performance: Efficient Use of Google\'s Geolocation API for High-Traffic Applications. Retrieved from https://dev.to/johnmiller/optimizing-performance-efficient-use-of-googles-geolocation-api-for-high-traffic-applications-50f [31] IF-SO. (2024). IP-to-Location vs. Geolocation API: Which Geotargeting Method is Best for You?. Retrieved from https://www.if-so.com/geotargeting-methods/ [32] Mozilla Developer Network. (n.d.). Using the Geolocation API. Retrieved from https://developer.mozilla.org/en-US/docs/Web/API/Geolocation_API/Using_the_Geolocation_API [33] softwareas.com. (2011). Cached Positions in the Geolocation API. Retrieved from https://softwareas.com/cached-positions-in-the-geolocation-api/ [34] Novu. (2025). The Ultimate Guide to Firebase Cloud Messaging (FCM). Retrieved from https://novu.co/blog/the-ultimate-guide-to-firebase-cloud-messaging-fcm/ [35] Firebase Blog. (2025). Making your FCM messages result in push notifications reliably on Android. Retrieved from https://firebase.blog/posts/2025/04/fcm-on-android [36] Firebase. (n.d.). Firebase Cloud Messaging. Retrieved from https://firebase.google.com/docs/cloud-messaging [37] JDIZM. (n.d.). node-express-firebase-mongodb. GitHub. Retrieved from https://github.com/JDIZM/node-express-backend-component [38] shanmugapriya1203. (n.d.). nexus-api. GitHub. Retrieved from https://github.com/shanmugapriya1203/nexus-api/ [39] India Meteorological Department, Government of India. (n.d.). Yearly Gridded Rainfall (0.25 × 0.25) data NetCDF File. Retrieved from https://www.imdpune.gov.in/cmpg/Griddata/Rainfall_25_NetCDF.html [40] NHESS Preprint, “Elevating Flash Flood Prediction Accuracy: A Synergistic Approach with PSO and GA Optimization,” preprint no. NHESS-2024-215, Natural Hazards and Earth System Sciences, 2024. [Online]. Available: https://nhess.copernicus.org/preprints/nhess-2024-215/nhess-2024-215.pdf. [41] Central Ground Water Board, “Pune District Maharashtra,” CGWB, Government of India, Tech. Rep., Nov. 2022. [Online]. Available: https://www.cgwb.gov.in/sites/default/files/2022-11/pune_f.pdf. [42] National Database for Emergency Management, “Landslide Atlas of India,” NDEM, Dept. of Space, Govt. of India, 2023. [Online]. Available: https://ndem.nrsc.gov.in/documents/downloads/Landslide%20Atlas_2023.pdf. [43] MapTiler Data, “Download Contour Lines Vector Tiles for Pune,” MapTiler, 2019. [Online]. Available: https://data.maptiler.com/downloads/dataset/contours/asia/india/pune/ [44] K. Halder, A. K. Srivastava, A. Ghosh, S. Das, S. Banerjee, S. C. Pal, U. Chatterjee, D. Bisai, F. Ewert, and T. Gaiser, “Improving landslide susceptibility prediction through ensemble Recursive Feature Elimination and meta-learning framework,” Sci. Rep., 2024. [Online]. Available: https://pubmed.ncbi.nlm.nih.gov/39939353/. [45] “Information gain ratio,” Wikipedia, The Free Encyclopedia. [Online]. Available: https://en.wikipedia.org/wiki/Information_gain_ratio. [46] NHESS Preprint, “Elevating Flash Flood Prediction Accuracy: A Synergistic Approach with PSO and GA Optimization,” preprint no. NHESS-2024-215, Natural Hazards and Earth System Sciences, 2024. [Online]. Available: https://nhess.copernicus.org/preprints/nhess-2024-215/nhess-2024-215.pdf. [47] S. Sharma, “Implement SVM and KNN for the flood prediction prb,” Kaggle Notebook, 2024. [Online]. Available: https://www.kaggle.com/code/shadesh/implement-svm-and-knn-for-the-flood-prediction-prb. [48] “Insights into landslide susceptibility: a comparative evaluation of multi-criteria analysis and machine learning techniques,” Arab. J. Geosci., 2025. [Online]. Available: https://www.tandfonline.com/doi/full/10.1080/19475705.2025.2471019. [49] AIScreen, “Real-Time Emergency Alert Display,” AIScreen Digital Signage Templates. [Online]. Available: https://www.aiscreen.io/digital-signage-templates/emergence-alerts/real-time-emergency-alert-display/.

Copyright

Copyright © 2025 Sankalp Mandlecha, Pranav Mane, Pranali Paigude, Darshan Patil. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.