TensorFlood| A Flood of Ideas

A Flood of Ideas

Your challenge is to develop a new methodology or algorithm that leverages Earth observation and critical infrastructure datasets to estimate damages to infrastructure caused by flooding. Make a measurable impact on the resilience of nations by helping the Earth observations community contribute to the United Nations’ primary effort to reduce disaster risk!

TensorFlood (Threat prediction of weather 'El Nino' in the peruvian coast)

Summary

TensorFlood is a powerfull smartphone-based application for flood damage prevention in the coast of Peru. It uses an algorithm to predict the level of threat of the damage caused by weather 'El Nino'. It helps people to prepare for this weather with helpful prevention guidelines and information of most affected cities.

How We Addressed This Challenge

What is it?

TensorFlood is a smartphone-based application that informs level of threat of weather 'El Nino' up to 9 months in the future for specific cities of the Peruvian coast. It also includes information about prevention guidelines, what to do, donation channels, and relevant aspects of the last weather 'El Nino'.

Why is it important?

It is important to provide an early warning system especially for areas of greater vulnerability, reduce risk factors and prepare an immediate response to the catastrophe. That is to say, avoid as much as possible to suffer material and economic damage, and above all save lives. Also be able to provide immediate information to different international and non-governmental institutions to help disaster areas.

What does it do?

The TensorFlood app has 4 critical functions:


  • Predict: It has a prediction algorithm that allows the user to get real-time predictions of level of threats of weather 'El Nino' up to nine months.
  • Inform: It provides information about previous tragedies caused by 'El Nino' and what were the damages, to get to know the real dangers.
  • Prevent: It has prevention manuals for lessening the damages caused by 'El Nino' and what is the workflow of the government to solve this problems.
  • Help: It allows a user to know the current donation channels to help the people affected during 'El Nino'


How does it work

It is an application that has the following sections.

1 Data: extracted data from different resources


  • Prediction: It alllows you to choose the city. Shows a plot of level of threat vs date time. It predicts up to nine months to the future.
  • Past: Shows a map of Peru with cities color coded. Each color represents the damage of each zone in previous 'El Nino' weather.

2 History


  • Shows the most important information of previous 'El Nino' weather.


3 How to Prepare


  • Prevention Guidelines.
  • What is the government approach.


4 Send Help


  • Donation channels.
  • What to donate.
  • Places where volunteer work is needed.

What do yo hope to achieve?

We want to provide an immediate information system to the population of civil society, public and private, for early warning and prevention of natural or man-made disasters. This will minimize material damage, reduce economic impacts and protect life.

Likewise, we want to achieve that, with this App, we can monitor the user's status by sending a signal that confirms to their relatives is okay, during and after the weather "El Nino".

How We Developed This Project

What inspired us?

In Peru, the weather 'El Nino' is usally asociated with destruction and loss. One of the direct consequences is Huayco, which is a peruvian term that refers to landslide. Some of the group members have been affected directly by the huaycos caused by floods and overflowing rivers, this experiences was very hard and we want to prevent future disgraces. We use an agile approach in order to try to make the minimum viable unit the more quickly as posible and to make iterations in order to improve it.

How did we make it?

The National Oceanic and Atmospheric Administration (NOAA) is the institution in charge of monitoring the temperature changes in the sea, specifically in Nino 3.4 zone, which are correlated to the impact of the weather 'El Nino'. They use historic daya of averaging 30 years of temperature and compared it to averaging 3 months of temperature to get and indicator called Oceanic Nino Index. The problem of this indicator is that it does not provide accurate information of which cities are going to be the most affected, so we proposed a solution.

We made an algorithm that predicts the level of threat of a specific city in Peru. We use historic data of temperature, precipitation level, wind vector, globar radiation, relative humidity and ONI to obtain a regression model that predicts from level threat 0 to 3 in the following months.

With this new indicator, people living in those cities or people near those cities, can see real-time prediction of the level threat of 'El Nino' in the following 9 months, which make it a powerfull tool for preparation and prevention. In the app we also include some prevention tips and guides, information regarding the last 'El Nino' and the principal casualties. At last we include a informative section to get to know donation channels, what can you donate and where can you provide volunteer services.

What problems and achievements did your team have?

We had some issues with the limitations of the data sources and the time we have had allocated, but we could overcome it and our best achievement was to do it on time.

How We Used Space Agency Data in This Project
  • We used Giovanni Earth Data to get images of Peru in flood seasons to obtain a better visualization of what are the most affected cities.
  • For the inputs in our model we used Meteomatics to get temperature, wind vector, relative humidity, precipitation level and global radiation.
Project Demo
Project Code
Data & Resources

Data sources




Tags
#machine learning, #floods, #deeplearning,#tensorflow, #El Nino
Judging
This project was submitted for consideration during the Space Apps Judging process.