ASAD SHABBIR PREddddddddddddSENTATN.pptx

ffgggg

1. ENHANCING FLOOD MONITORING AND PREVENTION THROUGH MACHINE LEARNING AND IoT INTEGRATION SUPERVISOR : DR IMRAN SHAFI PRESENTED BY : SYED ASAD SHABBIR

2. Introduction Literature Review Research Gaps Research Questions Research Objectives Proposed Methodology ML Algorithms Implementation Results Conclusion OUTLINE

3. Floods have become a significant problem in Pakistan, causing loss of life, property, and resources. The country is highly susceptible to heavy rainfall during the Monsoon season. Lack of readiness about upcoming floods makes the situation more severe. Developing a predictive system using Machine Learning and IoT can help mitigate the impact of floods and protect communities. INTRODUCTION

4. The occurrence of flood is influenced by the relationship between intensity of rainfall, how effectively water can drain away and how much water flows over the surface . In hydrology, this relationship can be represented by the rational method equation: Q = CiA Where : Q represents the flow rate . C is the runoff coefficient . i is the rainfall intensity, measured in units of depth per unit time (e.g., millimeters per hour). A is the drainage area contributing to the water flow surface INTRODUCTION

5. INTRODUCTION Top Level Domain Second Level Domain Third Level Domain

6. Water Station

7. Repeater Station

8. Siren Station

9. LITERATURE REVIEW M. Acosta- Coll , F. Ballester-Merelo, M. Martinez- Peiró , and E. De la Hoz -Franco, “Real-time early warning system design for pluvial flash floods—A review,” Sensors , vol. 18, no. 7, p. 2255, 2018 Title: Real-Time Early Warning System Design for Pluvial Flash Floods Outcome1 : Capability to provide timely and accurate W arnings Outcome2 : Significant reduction in property damage during flash floods.

10. LITERATURE REVIEW M. Acosta- Coll , F. Ballester-Merelo, M. Martinez- Peiró , and E. De la Hoz -Franco, “Real-time early warning system design for pluvial flash floods—A review,” Sensors , vol. 18, no. 7, p. 2255, 2018 Title: Early warning of impending flash food based on  AIoT Outcome1 : Continuous real-time data collection and analysis Outcome2 : Advanced AI techniques improve the accuracy and reliability of predictions.

11. LITERATURE REVIEW M. Acosta- Coll , F. Ballester-Merelo, M. Martinez- Peiró , and E. De la Hoz -Franco, “Real-time early warning system design for pluvial flash floods—A review,” Sensors , vol. 18, no. 7, p. 2255, 2018 Title: An Intelligent Early Flood Forecasting and Prediction Leveraging Machine and Deep Learning Algorithms with Advanced Alert System Outcome1 : LSTM and RNN models perform well in terms of MAPE Outcome2 : ANN shows the highest prediction accuracy.

12. LITERATURE REVIEW M. Acosta- Coll , F. Ballester-Merelo, M. Martinez- Peiró , and E. De la Hoz -Franco, “Real-time early warning system design for pluvial flash floods—A review,” Sensors , vol. 18, no. 7, p. 2255, 2018 Title: A Multi-Modal Wireless Sensor System for River Monitoring Outcome1 : Capability to monitor and transmit data in near-real-time Outcome2 : Accuracy of sensors in measuring river level, rainfall, and wind speed

13. Limited Datasets make Watershed Analysis S pecific Data collection essential for Flood Prediction. Radar-based level monitoring offers enhanced robustness due to its ability to accurately measure levels in challenging conditions such as harsh environments, where other methods may struggle to provide reliable measurements. RESEARCH GAPS

14. How can community-based flood warning systems leverage IoT technology to enhance early detection of flood alerts? What are the most suitable communication channels for delivering flood warnings to community members in real-time through IoT devices? How can machine learning algorithms be utilized to analyze sensor data from IoT devices for accurate flood prediction and warning generation? How can machine learning algorithms be trained and optimized using historical flood data and IoT sensor data to improve the accuracy and reliability of flood warning systems? RESEARCH QUESTIONS

15. L eads to collection of own dataset L eads to IoT installation at multiple sites L eads to data driven machine learning model L eads to choice of new sensors RESEARCH OBJECTIVES

16. PROPOSED METHODOLOGY Utilize IoT (Internet of Things) devices to collect data from various environmental sensors, including rainfall, river levels Integrate data streams from multiple sources to create a comprehensive and dynamic dataset Dataset Data Preprocessing Data Engineering Design of ML and DL Models Rainfall >50mm YES NO FLOOD FLOOD

17. DATASET

18. DATASET Preparation Installation of all required Libraries Load Dataset Data Preprocessing(Null Values Checking,Noise,Label Encoder) Data Splitting Ratio(70% for Training and 30% for Testing) Split features and Target(Rain_mm_total)

19. ML ALGORITHMS IMPLEMENTAION 1D CNN A 1D CNN for time series data is like a detector looking for patterns in a timeline. It's good at spotting trends or features in the data that help predict what might happen next.

20. ML ALGORITHMS IMPLEMENTAION Multi-Step LSTM LSTM is known for its ability to store and access information over long sequences. predicting several steps ahead in a time series data, like foreseeing the weather for the next few days based on past patterns

21. Results Model Performance Metric Mean Square Error 1D CNN 0.92 Multi-step LSTM 0.30

22. Comparative Analysis Study Methodology Algorithm Epoches Used Mean Square Error Proposed Study IoT and ML Multi-Step CNN 50 0.92 Proposed Study IoT and ML Multi-Step LSTM 50 0.30 Xilin Xia[11] DL CNN 50 0.005 Jian Sun [12] LSTM AND IoT LSTM 100 0.004

23. Conclusion In this study, we integrated Machine Learning (ML) and Internet of Things (IoT) technologies for flood monitoring and prevention. Using sensors to collect data on environmental parameters, we created a comprehensive dataset. We employed 1D Convolutional Neural Network (CNN) and Multistep Long Short-Term Memory (LSTM) models to predict Flood. The Multistep LSTM model achieved an MSE of 0.30, while the 1D CNN model had an MSE of 0.92, demonstrating the potential of ML and IoT in enhancing flood management and community resilience.

24. Achievement's Co-authorship on a Submitted Manuscript Title: "An Innovative IoT and GSM-based Flood Monitoring and Warning System for Mountainous Regions" Journal: Environmental Monitoring and Assessment Submission Date: February 22, 2024 Role: Co-author .

25. Achievement's Co-authorship on a Submitted Review Paper Title: "Review of Flood Monitoring and Prevention Approaches: A Data Analytic Perspective" Journal: Natural Hazards Submission Date: February 15, 2024 Role: Co-author

26. References [1] Chen J, Li Y, Zhang S. Fast Prediction of Urban Flooding Water Depth Based on CNN−LSTM.  Water . [2] Li, P.; Zhang, J.; Krebs, P. Prediction of Flow Based on a CNN-LSTM Combined Deep Learning Approach.  Water   2022 ,  14 , 993. [3] R. Tawalbeh, F. Alasali, Z. Ghanem, M. Alghazzawi, A. Abu-Raideh, and W. Holderbaum, “Innovative characterization and comparative analysis of water level sensors for enhanced early detection and warning of floods,” J. Low Power Electron. Appl. , vol. 13, no. 2, p. 26, 2023. [4] L. Mdegela , Y. De Bock, E. Municio, E. Luhanga , J. Leo, and E. Mannens , “A Multi-Modal Wireless Sensor System for River Monitoring: A Case for Kikuletwa River Floods in Tanzania,” Sensors , vol. 23, no. 8, p. 4055, 2023 [5] I . M. Hayder et al. , “An Intelligent Early Flood Forecasting and Prediction Leveraging Machine and Deep Learning Algorithms with Advanced Alert System,” Processes , vol. 11, no. 2, p. 481, 2023. [6] L. Mdegela , Y. De Bock, E. Municio, E. Luhanga , J. Leo, and E. Mannens , “A Multi-Modal Wireless Sensor System for River Monitoring: A Case for Kikuletwa River Floods in Tanzania,” Sensors , vol. 23, no. 8, p. 4055, 2023.

27. References [6] I. M. Hayder et al. , “An Intelligent Early Flood Forecasting and Prediction Leveraging Machine and Deep Learning Algorithms with Advanced Alert System,” Processes , vol. 11, no. 2, p. 481, 2023. [7] M. Acosta- Coll , F. Ballester-Merelo, M. Martinez- Peiró , and E. De la Hoz-Franco, “Real-time early warning system design for pluvial flash floods—A review,” Sensors , vol. 18, no. 7, p. 2255, 2018. [8] W.-T. Sung, I. V. Devi, and S.-J. Hsiao, “Early warning of impending flash flood based on AIoT,” EURASIP J. Wirel. Commun. Netw. , vol. 2022, no. 1, p. 15, 2022. [9] R. Tawalbeh, F. Alasali, Z. Ghanem, M. Alghazzawi, A. Abu-Raideh, and W. Holderbaum, “Innovative characterization and comparative analysis of water level sensors for enhanced early detection and warning of floods,” J. Low Power Electron. Appl. , vol. 13, no. 2, p. 26, 2023. [10] I . M. Hayder et al. , “An Intelligent Early Flood Forecasting and Prediction Leveraging Machine and Deep Learning Algorithms with Advanced Alert System,” Processes , vol. 11, no. 2, p. 481, 2023. [ 11] S . Kabir, S. Patidar , X. Xia, Q. Liang, J. Neal, and G. Pender, “A deep convolutional neural network model for rapid prediction of fluvial flood inundation,” J. Hydrol . , vol. 590, p. 125481, 2020 .

28. References [12] M . Yu, F. Xu, W. Hu, J. Sun, and G. Cervone, “Using Long Short-Term Memory (LSTM) and Internet of Things (IoT) for localized surface temperature forecasting in an urban environment,” IEEE Access , vol. 9, pp. 137406–137418, 2021.

29. LITERATURE REVIEW Title Methodology Dataset Main Findings Outcome Limitations A Multi-Modal Wireless Sensor System for River Monitoring Introducing a multi-modal, sensor-based, near-real-time river monitoring system that collects six parameters relevant to weather and river flood detection A multi-feature data set collected by a sensor-based river monitoring system for the Kikuletwa River Development of a sensor-based river monitoring system for collecting parameters for weather. River level (cm) Current hour rainfall (mm) Previous hour rainfall (mm/h) Wind speed (km/h) Ensuring the security and privacy of the collected data. L. Mdegela , Y. De Bock, E. Municio, E. Luhanga , J. Leo, and E. Mannens , “A Multi-Modal Wireless Sensor System for River Monitoring: A Case for Kikuletwa River Floods in Tanzania,” Sensors , vol. 23, no. 8, p. 4055, 2023

30. LITERATURE REVIEW Title Methodology Dataset Main Findings Outcomes Limitations Innovative Characterization and Comparative Analysis of Water Level Sensors for Enhanced Early Detection and Warning of Floods Testing four different water level sensors under various conditions, including different types of water, illumination, and analogue testing. Consists of four different water level sensors (two ultrasonic sensors, one IR sensor, and one pressure sensor). Accuracy and effectiveness of four different water level sensors (two ultrasonic sensors, one IR sensor, and one pressure sensor) in detecting water levels under various conditions, including pure and impure water Four water level sensors under different conditions and found that ultrasonic sensors and the IR sensor performed well for measuring distance to hard surfaces, while the pressure sensor was more suitable for water level measurements ( ability to accurately gauge water pressure, which directly correlates with water depth) The IR sensor may not be the best choice for water level measurements due to the nature of light and the need for additional data processing. - The pressure sensor could not be used in the same experiment as the ultrasonic sensors and IR sensor due to its working principle R. Tawalbeh, F. Alasali, Z. Ghanem, M. Alghazzawi, A. Abu-Raideh, and W. Holderbaum, “Innovative characterization and comparative analysis of water level sensors for enhanced early detection and warning of floods,” J. Low Power Electron. Appl. , vol. 13, no. 2, p. 26, 2023.