Welcome to the CML Research in Flood Forecasting Repository, a comprehensive collection of ongoing projects dedicated to developing Predictive Statistics, Machine Learning, Quantum Machine Learning algorithms in flood forecasting.
Natural disasters, particularly floods, pose a significant threat to lives, infrastructure, and economies worldwide. The increasing frequency and intensity of such events, driven by climate change and urbanization, underscore the urgent need for more accurate and reliable forecasting methods. Developing advanced techniques in predictive statistics, machine learning, and quantum machine learning offers the potential to better anticipate these events, providing critical time for preparation and response. By improving the precision of flood predictions, we can mitigate the devastating impacts of these disasters, safeguard communities, and enhance the effectiveness of early warning systems, ultimately contributing to greater resilience in the face of natural hazards.
This repository is organized based on the methods used to conduct the study, each exploring unique aspects of the models analyzed. Below, you will find a brief description of each method, along with links to their respective folders for more detailed information.
This project explores the use of predictive statistics for flood forecasting, with a primary focus on autoregressive modeling techniques. The first approach involves the application of methods such as Autoregressive Integrated Moving Average (ARIMA) and Seasonal ARIMA (SARIMA) to analyze and predict key hydrological variables, including water levels and discharge rates. By modeling the temporal dependencies and seasonality inherent in the time series data, these methods aim to produce accurate and reliable forecasts that can be crucial for early flood warning systems.
Ho, S.L. & Xie, M., 1998. Computers & industrial engineering, 35(1-2), pp.213-216.
You can check project updates here : PS Log
This project consists of two parts: the first part focuses on the application of K-Nearest Neighbors (KNN) algorithms for both classification and regression tasks, aiming to predict flood occurrences and quantify key hydrological variables such as water levels and discharge rates. By utilizing a time series dataset spanning several decades, the KNN models are trained to identify patterns and make forecasts with high accuracy. In the second part, the project advances to the development of Recurrent Neural Networks (RNNs), which are well-suited for sequential data, to further enhance the predictive capabilities.
Sherstinsky, A., 2020. Physica D: Nonlinear Phenomena, 404, p.132306.
You can check project updates here : CML Log
This project focuses on developing Quantum Machine Learning models to enhance flood prediction accuracy, with an initial emphasis on implementing Quantum K-Nearest Neighbors (QKNN) algorithms. By leveraging quantum computing's potential to process complex data patterns more efficiently than classical methods, the project aims to improve the predictive capabilities of traditional flood prediction models. The study involves analyzing large datasets of hydrological and meteorological parameters, applying QKNN to classify flood events, and comparing the performance with classical KNN models. The ultimate goal is to demonstrate the advantages of quantum approaches in real-world environmental forecasting applications.
You can check project updates here : QML Log
First you need to check if you have git software installed on your machine. You can check by typing the following in your terminal:
git --version
If no version appears, then you will need to download it. Depending on your operating system, you can install it as follows:
I. Mac OSX
- Install the package from https://git-scm.com/download/mac.
II. Linux
Depending on your distribution:
- Debian and Ubuntu-based distros:
sudo apt-get install git
- Red-Hat-based distros:
sudo yum install git
Once done, from a new terminal create a directory dedicated for this tool with the name QML_Flood_Prediction, and from it, clone this github files and intall the necessary components:
git clone https://github.com/GabrielWendell/QML_Flood_Prediction/tree/main
cd QML_Flood_Prediction
pip install -r requirements.txt
This will download the code and instructions.
├── Data_preprocessing/ # Folder for data preprocessing tasks
│ ├── Data/ # Raw data files related to data preprocessing
│ ├── Plots/ # Plots generated during data preprocessing
│ ├── Reduced_Data/ # Processed and reduced data files
│ ├── Data_Reduction.ipynb # Jupyter Notebook for data reduction
│ ├── CutiveMono-Regular.ttf # Custom font used in plots
│ ├── GillSans-Light.ttf # Custom font used in plots
│ ├── src/ # Source code for data preprocessing
│ └── README.md # Detailed README for this project
├── CML/ # Folder for classical machine learning analysis
│ ├── Notebooks/ # Jupyter Notebooks for ML models
│ │ ├── KNN_Model.ipynb # K-Nearest Neighbors model
│ │ └── RNN_Model.ipynb # Recurrent Neural Network model
│ ├── Plots/ # Plots generated during ML analysis
│ ├── src/ # Source code for machine learning analysis
│ └── README.md # Detailed README for this project
└── Predictive_Statistics/ # Folder for predictive statistics analysis
├── Notebooks/ # Jupyter Notebooks for time series models
│ ├── Correlation_Models.ipynb # Correlation model analysis
│ ├── Fourier_Analysis.ipynb # Fourier analysis
│ ├── ARIMA_SARIMA.ipynb # ARIMA and SARIMA models
│ ├── CutiveMono-Regular.ttf # Custom font used in plots
│ ├── GillSans-Light.ttf # Custom font used in plots
│ ├── README.md # Detailed README for this project
│ └── src/ # Source code for time series analysis
├── .gitignore # Git ignore file
├── LICENSE # License file
├── README.md # General README (this file)- The processed and treated data can be found in the
Reduced_Datafolder
We welcome contributions from the academic and scientific communities. If you are interested in collaborating or have any suggestions for improving the projects, please refer to the individual project folders for specific guidelines or reach out via the contact information provided below.
This repository is distributed under the GNU v.3.0 License. Please refer to the license file for more information.
For any inquiries related to this repository or ongoing research, please contact:
- Project Members
- GitHub
