Skip to content
@cropmosaiks

CropMOSAIKS

CropMOSAIKS: An Open Source Pipeline for Remote Sensing of Crop Yields: A Zambia Case Study logo

Purpose

This GitHub organization hosts the code and documentation used in a Capstone Project for the Masters of Environmental Data Science program at the Bren School of Environmental Science and Management, University of California Santa Barbara. This project intends to adapt, extend, and apply the MOSAIKS framework described in Rolf et al. 2021 to encode monthly imagery from the Landsat and Sentinel 2 satellite missions through random convolutional features over the years 2013 - 2021. The features are produced using an unsupervised machine learning approach called random kitchen sinks. The resulting features are task-agnostic and therefore can be used to model a wide variety of tasks. The features are geolocated and can be geospatially joined to observed data to answer a plethora of questions.

In this application of the MOSAIKS approach, the features are joined to crop yields in Zambia and a model is trained on this merged data using a supervised machine learning approach, cross validated ridge regression. Our model is optimized for this specific task through systematically comparing model perfomance for various temporal subsetts including month and year range, spectral band combinations, spatially masking for cropland, several interpolation approaches, and weighted averages for features summarized to district levels. The optimized model is then applied to predict crop yields for this region for subsequent years in which Zambia lacks crop yield data: 2020 - 2021. This project is limited by data availability and is restricted to the country of Zambia.

What is here?

This organization hosts three repositories including:

  1. Featurization
  2. Modeling
  3. images

The top two repositories have independent README's that provide detailed explanations of their notebooks, data sources, and file organization. The third repository includes graphics generated for presentations.

Contributing

The capstone project was completed on June 9th, 2022. Some team members will be continuing work in this field and the repositories are likely to stay active for some time to come. Suggestions for improvements to the code or documentation is welcome and encouraged. Please submit questions, comments, or code via issues or pull requests on either of the repositories. To correspond with the data scientists who produced these materials, please see their personal GitHub accounts listed below and feel free to contact them via email:

Grace Lewin
Cullen Molitor
Steven Cognac
Juliet Cohen

For rules and regulations for this organization, please see the Code of Conduct

Pinned Loading

  1. Featurization Featurization Public

    Satellite image encoding with random convolutional features

    Jupyter Notebook 1 2

  2. Modeling Modeling Public

    Training and applying crop yield models

    Jupyter Notebook 3

Repositories

Showing 5 of 5 repositories
  • crop-modeling Public
    cropmosaiks/crop-modeling’s past year of commit activity
    Jupyter Notebook 0 Apache-2.0 0 0 0 Updated Feb 2, 2024
  • .github Public
    cropmosaiks/.github’s past year of commit activity
    0 2 1 0 Updated Aug 23, 2022
  • images Public
    cropmosaiks/images’s past year of commit activity
    Jupyter Notebook 0 Apache-2.0 0 0 0 Updated Aug 23, 2022
  • Modeling Public

    Training and applying crop yield models

    cropmosaiks/Modeling’s past year of commit activity
    Jupyter Notebook 0 MIT 3 0 0 Updated Jun 3, 2022
  • Featurization Public

    Satellite image encoding with random convolutional features

    cropmosaiks/Featurization’s past year of commit activity
    Jupyter Notebook 1 MIT 2 0 0 Updated Jun 3, 2022

Top languages

Loading…

Most used topics

Loading…