Sustainable Disaster Response Alert Mechanism (SD-RAM) project developed by Team Horizon BD is indeed an impressive and comprehensive effort to address the challenges posed by natural disasters, with a particular focus on lightning strikes and floods in Bangladesh. Here are some key points to highlight:

Lightning Prediction Model : In order to predict thunderstorms we have created a Deep neural Network Model that can predict thunderstorm at 95% accuracy.

As feature to predict thunderstorms we have

T2M MERRA-2 Temperature at 2 Meters (C)
T2MDEW MERRA-2 Dew/Frost Point at 2 Meters (C)
T2MWET MERRA-2 Wet Bulb Temperature at 2 Meters (C)
QV2M MERRA-2 Specific Humidity at 2 Meters (g/kg)
RH2M MERRA-2 Relative Humidity at 2 Meters (%)
PRECTOTCORR MERRA-2 Precipitation Corrected (mm/hour)
PS MERRA-2 Surface Pressure (kPa)
WS10M MERRA-2 Wind Speed at 10 Meters (m/s)
WD10M MERRA-2 Wind Direction at 10 Meters (Degrees)
WS50M MERRA-2 Wind Speed at 50 Meters (m/s)
WD50M MERRA-2 Wind Direction at 50 Meters (Degrees)

One of the main challenges was to find and collect clean and relevant datasets across parameters in the same time frame. The dangerous lightning strikes (encoded as 1) represented only 0.03%, leading to a biased dataset.

To combat this, we have used SMOTE model

When you have an imbalanced dataset, where one class (the minority class) has significantly fewer examples than the other class (the majority class), traditional machine learning algorithms may perform poorly because they tend to be biased towards the majority class. SMOTE is a method designed to mitigate this issue by oversampling the minority class. Here's how it works:

SMOTE selects an example from the minority class.
It then selects k-nearest neighbors (typically, k is set to 5) of that example from the same class.
SMOTE generates synthetic examples by interpolating between the selected example and its k-nearest neighbors.
These synthetic examples are added to the dataset, effectively increasing the size of the minority class.

For a specific region the information (feature for our model) seamlessly flows into our artificial deep neural network model, which is hosted on a remote server, allowing us to predict lightning probability with remarkable accuracy and precision.

Portable Lightning Arrester: The portable lightning arrester is a unique feature that can provide personal protection to users, especially in areas lacking traditional lightning protection infrastructure. This could potentially save lives during lightning storms.

Innovative Communication Device: The "black-box" walkie-talkie device with long-range communication capabilities is crucial for maintaining communication during disasters when traditional networks are often disrupted. This can greatly aid coordination and rescue efforts.

Flood Prediction System: Given the recurring nature of floods in Bangladesh, a real-time flood prediction system is invaluable. It enables users to make informed decisions regarding evacuation and preparedness, potentially minimizing casualties and property damage.

Natural Event Tracker: Leveraging NASA's data and converting it into user-friendly maps can help raise public awareness about various natural events, enabling better preparation and response to a wide range of disasters beyond lightning and floods.

User-Generated Safe House Locations: Encouraging users to contribute safe house locations fosters community engagement and a sense of collective responsibility. This collaborative approach enhances disaster resilience and community preparedness.

The project addresses critical aspects of disaster management, ranging from prediction and protection to communication and community involvement. By doing so, it not only saves lives but also minimizes the damage caused by natural disasters, ultimately contributing to the safety and well-being of the population in Bangladesh and potentially serving as a model for disaster response in other regions as well. This multidimensional approach to disaster management is commendable and can have a significant positive impact

2023 NASA Space Apps Challenge

Horizon BD