SkyWolves

StonyBrookWeather.com
StonyBrookWeather.com

StonyBrookWeather.com

Team SkyWolves Feb 22, 2025 Team members: Hajun Shin, Aninda Saprotiv Roy, Broja Brota Saha, Ajmain Faieq-Ul Hoque

Introduction There are many kinds of weather forecasting providers in the world. But all of them can not be accurate all the time. Some are good at predicting snow, and some others can predict temperature well. In this prospect, we want to aggregate them in one and share it to everyone. That’s why we planned this website.

Main Idea The backend AI is calculating weights that will be multiplied to each of the weather forecasting providers’ predictions by each of their accuracy in the sections of predictions of snowing, raining, and temperature. Then the backend AI is expected to dynamically emphasize more accurate data within each section by reinforcing the weight of higher-accuracy values, thereby introducing a bias toward more reliable predictions.

Progress We planned to use the api of OpenWeatherMap, WeatherBit, NationalWeatherService(NOAA), which are providing weather prediction api for free or at almost low prices. We decided to use Netlify for the frontend server and Heroku for the backend server. The Heroku-hosted backend server retrieves next-hour weather forecasts from various APIs every hour, compares the previous hour’s predicted data with the actual current weather to evaluate accuracy, and then calculates and continuously updates weights in real time. Additionally, every midnight, the server fetches one-week weather forecast data from each API, multiplies these forecasts by the calculated weights to generate our proprietary integrated weather prediction, and updates this information on the backend for display on the frontend. We also added switches that can change the units of Temperature between Fahrenheit and Celcius for different cultures.

Challenges In this project, we encountered significant challenges related to the ongoing costs of maintaining a complex backend infrastructure, as well as the expenses associated with using multiple weather forecasting APIs. The financial constraints imposed by these maintenance costs have pushed us to explore more cost-effective, scalable solutions. Although our initial deployment focused on serving the Stony Brook area, we have ambitious plans to expand our service to cover weather forecasting on a global scale. This global expansion will not only require us to optimize our backend infrastructure but also to negotiate better terms or identify alternative API providers that can deliver high-quality weather data at a lower cost. By addressing these challenges, we aim to provide a reliable, real-time weather forecasting service that is accessible to users around the world, ensuring both technical excellence and financial sustainability.