High-Level Project Summary
Climate change is one of the biggest problems facing humanity today. It's up to us, young scientists, engineers, and creatives, to address this challenge. We present to you HeatML, a desktop application with the sole purpose of warning civilians and governments about heat-related events and providing instructions towards mitigating them. We provide a simple interface with relevant weather data and important forecasts. We provide our own API and data submission platform for crowd-sourcing data collection. Most importantly, we guide users in protecting themselves, loved ones, and their communities.
Link to Project "Demo"
Link to Final Project
Detailed Project Description
What does it do?
HeatML is a desktop application that provides users with an interface to view heat-related weather data over time for a given location. We use time series plots to show heat index, forest fire index, soil moisture deficit, and harmful gas concentrations (CO, NO2, and O3) over time. We also predict future values to be used in the warning message. The warning message outlines danger levels for heat stress, droughts, forest fires, and air pollution based on the forecasted values. We hope to provide further customization to the warning message using user health data which is provided on sign-up.
We also created a UI for data submission and an API helper. We plan to complete the functionality of these pages in the future. The API helper will provide a custom URL based on chosen data criteria. Curious students and researchers will be able to access any of our data through an API. The data submission page gives an opportunity to outsource data collection to users. This would improve the quantity of data that's available, resulting in better analysis.
How does it work?
We used the Meteomatics API to obtain all the historic weather-related data displayed on the app. We implemented the Stata ARIMA time series estimation algorithm in a Jupyter notebook (not implemented in the app) to forecast future values for our heat index time series plot. We plan to implement the algorithm for the other data in the future. A plot of the forecasting is shown below.
At the moment, the app uses a linear regression algorithm to estimate future values, but the Stata ARIMA algorithm will be added to the app in the future.
To create the warning message, we use known threshold values discovered in our research which allows us to determine danger levels.
What benefits does it have?
Users will be able to access heat-related disaster data and predictions that they can use to make informed decisions. For example, our warning system may alert them of an upcoming risk of heat stress. Even if they don't know how to prepare for this event, our software provides suggestions as to how they may protect themselves.
Tools and Software
We developed the application using Python, Flask, HTML, CSS, and Javascript. We used the Python web2gui library to convert the Flask website to a GUI desktop app. All data analysis was done in Jupyter notebooks using the pandas library. We also used scipy and statsmodels to implement our machine learning algorithms. All version control was done using Git/GitHub. We used Discord for communication.
Space Agency Data
We were inspired by the CSA data collected by OSIRIS (Canada's Optical Spectrograph and Infrared Imaging System) which is the payload on Sweden's Odin satellite. The satellite collected atmospheric composition data at many altitudes and locations over time. Through an analysis of the data using Python's pandas library, we determined that the concentration of greenhouse gases has increased over time. Aerosol measurements at varying altitudes can be seen in the graphs below.
As you can see, the coefficient seems to have increased over time. A similar analysis was done for other greenhouse gases as well. As a result, we decided that we wanted to also tackle air pollution with our project. We decided to focus on the human health aspect of it and how we can mitigate the damages to our health by unsafe concentrations of these harmful gases.
Hackathon Journey
We loved this challenge! We learned so much about Space, data collection, data processing, and software development. We also learned a ton about the environment and the effects of climate change.
A major setback we had was our inexperience with data collection and processing. We struggled to find relevant data to display on the app. To resolve this, we used the Meteomatics API to collect data about any location. It was an immense help in finishing the project.
References
Here are all the resources we used:
- https://www.meteomatics.com/en/weather-api/
- https://ehs.research.uiowa.edu/occupational/heat-stress
- https://ambientweather.com/heind.html
- https://www.weather.gov/ama/heatindex
- https://www.mayoclinic.org/diseases-conditions/heat-exhaustion/symptoms-causes/syc-20373250
- https://www.cdc.gov/niosh/topics/heatstress/recommendations.html
- https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2900329/
- https://www.ishn.com/articles/106977-pre-existing-conditions-can-worsen-heat-exposure-effects
- https://scied.ucar.edu/learning-zone/air-quality/how-weather-affects-air-quality
- https://donnees-data.asc-csa.gc.ca/en/dataset/6c3f7e9c-7b43-4a6b-a924-a1bef1a9cf74
- https://www.c2es.org/content/heat-waves-and-climate-change
- https://www.c2es.org/content/heat-waves-and-climate-change/
- https://www.kaggle.com/cdminix/us-drought-meteorological-data
- https://www.kaggle.com/elikplim/forest-fires-data-set
- https://medium.com/@llmkhoa511/time-series-analysis-and-weather-forecast-in-python-e80b664c7f71
- https://www.youtube.com/watch?v=PfVxFV1ZPnk
Tags
#heat, #meteomatics, #data, #api, #air pollution, #heat stress, #droughts, #forest fires
Global Judging
This project has been submitted for consideration during the Judging process.