Flood Damage Control: Estimating damage with GIS and ML

Inspiration

Based on the prompts given to us, we brainstormed a variety of ideas to benefit the community. One concept that resonated with us was a disaster alert and preparation system, with a specific focus on flood alerts and their impact on farms.

What it does

Our application allows you to input your farm's parameters, such as coordinates. Using this data, we aim to predict potential flood events, the extent of damage they might cause to your farm, and the resulting profit loss.

How we built it

Flood Predictive Model: We first built a model to predict flood occurrence and potential locations.
Data Collection: We then collected a variety of Normalized Difference Vegetation Index (NDVI) data from farms, both before and after flood events.
Machine Learning Integration: This NDVI data was fed into a machine learning model to predict the future NDVI.
Profit Loss Estimation: We used the difference between the current and predicted future NDVI values, along with estimates of crop loss and market prices, to predict potential profit loss from a flooded field.

Challenges we ran into

One of the main challenges we encountered was data collection. We needed a variety of NDVI data from farms, both before and after flood events. Overlaying flood data with satellite data, particularly Landsat data, can be difficult because satellites don't always cover specific areas, cloud cover can interfere with data collection, and other factors can introduce noise into the data, such as mountains appearing more green after heavy rains.

Accomplishments that we're proud of

We are proud of developing a comprehensive flood prediction model to estimate flood-affected areas within a farm and initiating a machine learning model.

What we learned

For some of us, this was our first time working with geographic information systems (GIS). The software was helpful in teaching us how to collect geographical data and the flood predictive model we worked on was insightful in demonstrating the many features involved in geographical predicting. These features include factors like plot yield, soil erosion caused by rain, pest infestation, and a variety of others that we will need to consider for future implementation.

What's next for Flood impact analysis

Our next step is to expand our application to include predictive price calculations for crops, considering future market trends rather than just the current market price. This will provide farmers with a more accurate estimate of potential profit loss.