A machine learning software designed to model soil cycles and recommend amendments to improve carbon sequestration and crop yield

Photo of Karina Andersen
4 3

Written by

Additional categories (optional)

  • Technology

Are you employed by T-Mobile or related to an employee of T-Mobile?

  • No

Eligibility: Date of Birth

March 8, 2003

Help us stay in touch!

Address: 5565 Cabrillo Ct, Rocklin, CA 95765 Phone: (916) 581-2768

Date You Started Your Project Started


Project Stage: Select the description below that best applies to your approach.

  • Idea (hoping to get started in the future)

1. The Problem: What problem are you helping to solve?

Over the past century, damaging agricultural practices have significantly reduced soil quality, depleting crop nutrients and releasing carbon dioxide into the atmosphere, contributing to global warming. Additionally, with the world population on the rise for the foreseeable future, food demand will continue to increase. Therefore, it is vital that farmers work to improve soil quality to encourage more nutritious and climate-friendly crop yields.

2. Your Solution: How are you planning to solve this problem? Share your specific approach.

SoilHorizons is a unique soil modeling software that will have the functionality to apply a set of quantitative and qualitative parameters (determined from a simple soil test taken from the plot being modelled) to a series of machine learning based algorithms and predict soil response to the application of different amendments, outputting the optimal amendment for crop yield, soil health, and carbon sequestration. The program will be written with C++ due to its versatility in incorporating both high and low level languages and object-oriented programming. The algorithms will be trained by semi-supervised learning in order to both deduce patterns and optimize predictions using training data which will be sourced from the public data repository provided by the International Soil Modeling Consortium. The model will take into account the dynamic components of soil ecosystems, the lack of which is a pitfall for many soil models. From the modeled processes and cycles, it will then be able to predict the effectivity of applying certain amendments to the soil and determine the optimal amendment for the given environment.

3. Personal Journey: What’s the story behind why you decided to start this project?

My interest in applying scientific innovation to fight global warming began with a project in eighth grade assigned by my English teacher, for which I prototyped a device that converted carbon dioxide emissions from car exhaust into calcium carbonate. Having little success with such an idea, I brainstormed many different inventions in my freshman and sophomore years before stumbling upon the potential of soil while creating a documentary about the effects of agriculture on climate change. After a brief internship with the Working Lands Innovation Center at UC Davis this past summer, a group studying the ability of different amendments to increase soil carbon sequestration, coupled with attending the summer program AI4ALL after my freshman year, I began to consider how I could use machine learning to take the guess-work out of soil amendments and SoilHorizons was born.

4. Selfie Elevator Pitch: Include 1-minute video that answers the following “I am stepping up to make change because...”

5. Example: Please walk us through a specific example of what happens when a person or group gets involved with your project.

Over the past decade, Farmer Fred has noticed that his yearly soil tests have repeatedly returned imbalanced levels of critical and trace nutrients despite applying the standard fertilizers recommended by his soil testing company. Fred is worried that these imbalances are hurting his yield, which has been steadily decreasing. Frustrated, Fred contacts SoilHorizons and follows the step-by-step guide to collect soil samples from his fields and take simple measurements of his croplands. The data is inputted into the program, which generates a model of his soil ecosystem and predicts an optimal amendment for his soil. An order is placed for the creation of such amendment, and Fred applies it to his cropland in the dosage recommended by the software. After continuing soil testing, Fred notices that his soil health has drastically improved, thereby increasing his production and profit.

6. The X Factor: What is different about your project compared to other programs or solutions already out there?

The field of soil modeling, while fairly robust, still requires significant development and will continue to evolve greatly. Additionally, while soil testing recommendations offered by small corporations and universities have reasonable success in improving soil quality, the AI algorithms included in SoilHorizons would offer the ability to tailor amendment recommendations even more optimally to a soil, improving yields to a vastly greater degree. Additionally, very few soil models and soil tests include carbon sequestration as a factor in soil health, an aspect offered by SoilHorizons.

7. Impact: How has your project made a difference so far?

After its primary year of testing and development, I intend to expand SoilHorizons to farmers statewide (and later, if successful, nationwide), thereby improving agricultural soils and yields which may be reflected in increased economic output from the sector. At the same time, with the inclusion of carbon sequestration properties in SoilHorizons’ algorithms, this will also increase the rate at which carbon dioxide is removed from the atmosphere, returning it to its natural state in the land carbon sink and hopefully assisting in combating global warming. Throughout this entire process, I hope to make my journey in the development of SoilHorizons as transparent as possible, perhaps through a blog, and encourage greater public awareness of the impacts of agriculture and food supply on climate change.

8. What’s Next: What are your ideas for taking your project to the next level?

By the end of 2019, I hope to finish the first software prototype and begin small field testing over the winter. Once I can get a set of primary results, I will continue modifying, improving, and retesting the software until it reaches a level of effectiveness which would be suitable for greater public use. While I already have some connections to farmers in my local area, I am trying to forge even more for larger field testing when the software reaches such a point. By June of 2020, I hope to have sourced enough funding and partnerships through grants and competitions to launch as a true business and apply for a patent, which would hopefully encourage more farmers to trial the software on their lands.

9. Which of the following types of expertise would be most useful for you?

  • Project Plan & Strategy

10. Finances: If applicable, have you mobilized any of the following resources so far?

  • Friend support
  • Family support
  • Mentors/advisors

Help Us Support Diversity! Part 1 [optional] Which of the following categories do you identify with?

  • White (for example: German, Irish, English, Italian, Polish, French) (6)

Help Us Support Diversity! Part 2 [optional] Do you identify as part of any of the following underrepresented communities?

  • No, I do not identify with an underrepresented community

How did you hear about this challenge?

  • Recommended by others

Referral: If you discovered the Challenge thanks to an organization or person other than Ashoka or T-Mobile, who was it?



Join the conversation:

Photo of Lynnea Shuck

Karina, your idea of coming up with a tool that aligns agricultural output with carbon sequestration is really inspiring! I can imagine this being a great educational tool as well, such as in school gardens/in environmental science classes. Perhaps this might be an additional area for expansion!

Photo of Karina Andersen

Thanks, Lynnea! I'd love to find some way to incorporate this into an educational curriculum and may apply it next year if our school decides to start an organic garden.

View all comments