Earth Day 2023

Data Science to Combat Global Heating

Hackathon

Hackathon results

The winner of the Earth Day 2023 Hackathon is:

  • Tobias Kallehauge

Runners ups:

  • Martin Voigt Vejling

  • Mikkel Mandrup Nielsen

Repository

The files needed for the hackathon are available in the repository: github.com/everval/EarthDay2023

Instructions

The NetCDF file format

In this hackathon you are going to familiarize yourself with the NetCDF file format and use it to generate temperature predictions.

NetCDF (Network Common Data Form) is designed to facilitate access to array-oriented scientific data. NetCDF is a portable, self-describing format. This means that there is a header which describes the layout of the rest of the file, in particular the data arrays, as well as arbitrary file metadata in the form of name/value attributes. NetCDF is used extensively in the atmospheric and oceanographic communities to store variables, such as temperature, pressure, wind speed, and wave height. It is also used widely in climate modeling and analysis.

For this hackathon, you will be working on a subsample from the GISS Surface Temperature Analysis version 4 (GISTEMP v4). GISTEMP provides a measure of the changing global surface temperature with monthly resolution since 1880. It takes historical temperature data from land-based weather stations as input and combines these data to produce an estimate of temperature change over large regions.

Due to its compressing, in all programming languages opening NetCDF requires installing a package and a specific list of commands.

Data

The repository contains three data files in the NetCDF format with suffix .nc. The files contain data on temperature anomalies for a list of latitudes and longitudes across the globe in a 10 by 10 grid. In all files, the last year (12 observations) at the (6,6) grid is missing. Moreover:

  • Ex2 also has missing data in grids (5:6)X(6:7).

  • Ex3 has missing data in grids (5:7)X(5:7).

The task

Your goal is to impute/predict/forecast the missing observations at the (6,6) grid. You can use any method and programming language for prediction, but your results should be shown in a CSV (comma-separated values) file.

Alongside your predictions, the code used to create the predictions should be provided.

The winner of the hackathon is the person/team which predictions achieve the minimum squared loss from the true temperature anomalies while using the least amount of information.

We use a point system to determine the winner. Points are awarded in a reverse-order to the top 10 results from solving the exercise using the first dataset temp_anomaly_ex1.nc. That is, the top solution gets 10 points, second gets 9, and so on.

Once you are confident with your solution to the first exercise, additional points can be achieved by solving the problem using the second or third datasets. Points are awarded in a reverse order for the top 3 solutions for the exercise using the third dataset, and for the top 2 solutions for the exercise using the second dataset.

The format

You will have 3.5 hours for the task. You can work in groups or by yourself. We target the groups to be at a maximum of 3 participants, but bigger groups can be accommodated.

There is some example starter code using Julia in this repository, you should read it first to familiarize yourselves with the file format.

You may ask questions during the hackathon, but try to google things first to speed things up.