O17 Water: Enhanced AI Flood Map System
In the United States, many people struggle with extreme flooding conditions due to Major storms and elevating oceans as climate change continues to alter the geography of coastal and watershed. FEMA our emergency responding service has struggled to maintain accurate and reliable flood maps. FEMA does not anticipate trends in future flooding and is not able to efficiently warn people who are now in flood zones due to rising sea levels.
Our team consists of two siblings, Isha Patel and Ishan Patel.
Isha is a senior at Wooster High School and is deeply driven by creating realistic solutions to help others. Moreover, she is driven by the thoughts of improving our society. Isha had formed the platform for the solution and continues to develop and enhance it so that it can reach its full potential, by comprehending the feedback given by our mentors and applying it. Moreover, she prepares the gist of the pitch and develops on it the farther she goes.
Ishan is a sophomore at Danbury High School and is driven by finding the problem as to why something is not working the way it is supposed to be. Ishan plays an important role as the problem identifier and continues to document the new research that is found in order to incorporate it into our pitch.
In the field of water mapping, Geological Survey’s National Water Information System is similar to our app-based smartphone platform. Geological Survey’s National Water Information System is an interactive tool that maps water resources data at over 1.5 million sites across the country. The search tool allows the user to find sites by street address, location name, site number, state/territory, and watershed region, etc.
Our initial look at the project started out with identifying the problem. The problem we saw was that during a flood-related disaster, it takes our emergency responding systems too much time to analyze the disaster and the impact it had or could possibly have going forward. Our current solution to this problem ended up becoming an efficient method of comprehending and analyzing data. Ultimately, this data could be implemented at a local level.
Our initial solution consisted of similar variables to our current solution. In the beginning, we thought of an app that could record water lines and flood maps(and was oriented around the concept of flood maps). Now, however, our solution is oriented around the idea of emergency responding systems not being able to collect the necessary data on time. This significantly changes the outlook on our projects because now our project isn't about flood maps, it is based on the idea that rapid responses can save lives.
When discussing our initial solution to the problem, we found that it wasn't a viable option because we were able to find products that were very similar to ours and so that wasn't very unique. We found our specialty to reside in providing real-time data where that could be provided to emergency responders and they could quickly act upon anyone who is in danger.
One aspect of our solution that we found work really well is machine learning. We found it to be cost-effective and crowdsourcing as well, was able to collect data on a larger scale. Moreover, our solution is objective. Meaning that no one really influences that data and it is what is shown and what can be analyzed from it.
In the future, as our project develops and starts to take shape, we plan on partnering up with local emergency and rapid response teams. This way, we will be able to put our solution to the test and see how effective it is on the battlefield. We will be keeping a close watch on the efficiency of it to fix/tweak any necessary errors.
Our research consisted of looking up flood damages and reviewing articles written by those who were affected by them. Moreover, we read articles from organizations who were trying to solve a similar problem to ours. Inspiration from them, statistics of flood damages, and articles written by people who were affected all contributed to the formation of our solution.
Crowdsourced smartphone photos taken of their local neighborhood and environment.
- Identify water lines, measure water levels
- Geolocation abilities (coordinates, time)
Smartphone-based volunteered geographic information (VGI) platform.
Machine learning used to detect the flooding line of the classified image.
Random Forest Classification
Identifying key points (road lines, traffic lights)
* Calibration (Expected vs actual outcome)
Week 2 Pitch Presentation: https://docs.google.com/presentation/d/1yBdaorbYTar4uvsIKcCkG0cF5VI_1zO_gXSW66AWFOM/edit#slide=id.p
Week 3 Pitch Presentation: https://docs.google.com/presentation/d/1y6im2ZHjMT-X_gnQdjhw47l2GLn2oIA5Jd8jrRkMGPQ/edit#slide=id.p
Pitch 4 Pitch Presentation: https://docs.google.com/presentation/d/1cRXwypLnEUhOXwrWZNplEVwb3tCyv3WFozgd3IOeyRI/edit#slide=id.p
1 MIN VIDEO SUMMARY
The advice given by our mentor Josephine was to identify our stakeholders. Knowing that a stakeholder is a party that has an interest in a company and that primary stakeholders in a typical corporation are its investors, employees, customers, and suppliers, we moved on to brainstorming some stakeholders. Our list consisted of various stakeholders but was eventually boiled down to two main ones. The first one was local meteorological companies and the second one was rapid responders and first responders.
Our second mentor feedback came from Will and Amudha. They pointed out that our presentation was very straightforward and to the point, which was not a bad thing. They advised us to make it personal and include a story to emphasize the problem and to make the listeners more attentive. They also advised us to add visuals because it's more important to show than tell.