FloodWatch

🤼 Team

Crina - Third year computer science student at University of Twente from Moldova. Passionate about data science, machine learning and software development. Participating at my first ever hackathon.

Novo - Fourth year computer science student at University of Twente from Bangladesh. Aspiring Machine Learning Engineer. Participating in his first ever hackathon ever.

Ion - Third year computer science student at University of Twente from Moldova. Passionate about web development, co-founder of small web platform startup Liberti. This is my third hackathon event.

Eugen - Third year computer science student at University of Twente from Moldova. Interested in all things computer science: data analysis, strategist, planning and business perspective. This is my first hackathon event.

Lucian - Third year computer science student at University of Twente and graduate in Molecular Biotechnology from Babes-Bolyai University. Interested in keeping the team together as a project coordinator, planner, business strategist, and team mediator. This is my first hackathon event.

Alex - Former linguistics university student. Interested in the business aspect of projects and project coordination, with a focus on the migration part of the project, due to my background and experience in migration as a refugee integration teacher. This is my 2nd hackathon event.

💎 Idea

Our goal is to create a tool that helps medium-long term prediction and preparation of zones prone to flooding and reduce the number of deaths related and the number of people suffering a loss of life quality. We want to concentrate on floods in particular, because, due to global warming, lots of land will be underwater in the near future. Countries that have extensive coastlines are at risk of periodic widespread damage. People will be forced to migrate and leave their homes. The project tackles the sustainable development goals SDGs Climate Action and Life on Land.

Every life matters, which is why our solution can be applied to various regions in the world that struggle with water flashing from nearby hills, heavy local rainfall, coastal flooding and bursting of river banks. 

We want to help identify areas that are at risk of flooding in the future, and areas that are adequate for building more sustainable houses for the future. To decide if a country has flooding areas, we take the average altitude of that country and the threshold that once it is reached, flooding happens. The threshold is different for each country and depends on the average altitude. If the amount of precipitation goes over the threshold, we can decide that flooding might happen. 

The stakeholders are the general public, NGOs, humanitarian workers, and local authorities. Due to the nature of the project, we believe it could help the above stakeholders to navigate across flooded regions in all parts of the world, due to our use of satellite data. This is especially helpful for NGOs and local humanitarian workers responding after flooding catastrophes.

⛑ Space for International Development & Humanitarian Aid

Our project proposal is regarding the third challenge: forecasting and countering forced migration. The rising sea levels in general combined with the particular climate appearances, contribute to an environment that is dominated by changes in rainfall patterns and the disappearance of fruitful land in rural areas. With the heavy rainfalls that happen mostly in the rural parts of the country, people have to leave their homes and move to the cities in the center of the country, where elevation allows the existence of infrastructure of big cities. As a result, residents in these vulnerable areas are often forced to migrate to ensure their safety. Satellite data can help us predict and prevent this by informing NGOs and local governments to plan in advance, and to evacuate or prepare in advance for floods.

GLOFAS and EFAS are services available to members of organizations and businesses, or to members of the platform. Even though these services exist, people are still dying. We want to make our product available and easily accessible to the general public. This means that no account is necessary to view any of the services provided. Floods are shown in real time, which helps making economic and emergency decisions. Short-medium term predictions: 14-30 days, citizens can plan in advance of a possible flood. It can help people make better decisions when it comes to selling a property/buying it.

We are aware that with this project and our motivation to help solve forced migration induced by floods, we directly express the hierarchy between the ones that are capable of giving help and the ones that are in need of help. We are conscious about our positions in the global North and Global South, and we promise that our goal with this project is not to deepen inequalities or hinder attempts of countries to be independent and help themselves but to truly be innovative and find sustainable solutions for the future, for all of us. 

🛰 EU space technologies

We  use data such as: precipitation levels and optical satellite imagery data.  We implemented a web-based application which uses two machine-learning models to display both flooding data (detection of floods based on satellite imagery) and the probability of short-term flood (future) data visually. For prediction we use an ARIMA-model to do time series forecasting, using historical precipitation data. For the ARIMA model we use data from the cds.climate.copernicus.eu: Precipitation monthly and daily gridded data from 1979 to present derived from satellite measurements. 

Our detection model was trained on a dataset of labeled flood images from the US state of Louisiana, but we intend to use data from the Sentinel 2 from the Copernicus system of satellites. The forecasting model uses training data based on Dutch weather data, but in the future will be trained and tested on more Sentinel 1 data, which would allow us to expand the region of availability for forecasting, along with more precise short-term predictions.

In short, Floodwatchers is a comprehensive data-driven solution designed to predict and monitor floods by harnessing meteorological data from the Sentinel-1 satellite within the Copernicus system, as well as aerial images from the Sentinel-2 satellite, all while providing a user-friendly web application for real-time flood information accessible to the public and relevant authorities.

Explaining the architecture

• Data pulled from satellite to our server.

• Server processes the data and presents it to our publicly accessible web app.

Key Features of the Web Application:

• Real-time flood alerts and warnings.

• Historical flood event tracking and analysis.

• Customizable map layers showing satellite imagery, flood predictions, and flood detection results.

• User-friendly search and navigation features.

• Responsive design for accessibility on various devices.

Repository:

https://github.com/novojitsaha/cassini2023

https://github.com/wiren301/nl_coppernicus_hackathon