ellaampy/CocoaMiningDS

Introduction

Agriculture remains a cornerstone of Ghana's economy. More than half of the land area is dedicated to agriculture, and its expansion continues to accelerate. Despite this growth, the sector faces compounding challenges, including climate-induced risks such as prolonged droughts, which aggravate low yields and threaten long-term productivity. A dimension of Ghana's agricultural landscape is agro-ecological regions, which not only sustain primary agricultural activities such as cocoa production but also coincidentally overlap with areas rich in mineral resources where several extractive sector activities, such as gold mining, occur on small and large scales. This convergence has driven a transition from agriculture landuse to mining, reshaping economic priorities and landuse dynamics. Artisanal and small-scale gold mining (ASGM) is on the rise, expanding into cocoa-dominated landscapes and posing serious threats to agricultural soils, water resources, and cropland due to its unregulated and unsustainable nature. Leveraging satellite imagery and geospatial analysis, we mapped cocoa and mining landuse, revealed areas where ASGM has taken over cocoa farmland, assessed the economic value of the transitioned areas, analyzed atmospheric elemental mercury in the cocoa landscape and developed a cocoa risk map.

Study area

Data

• CocoaMiningDS data contains a 3-class label (background-1, mining-2, cocoa-3), with dimensions 1 x 128 x 128

• CocoaMiningDS relies on satellite imagery from SmallMinesDS which contains 4270 patches; 2175 each for 2016 and 2022

• The corresponding code for segmenting cocoa and mining landuse can be found here

Citation

If you use the dataset or supporting code in your research, please cite SmallMinesDS and "Cocoa under Threat" as :

@ARTICaLE{10982207,
  author={Ofori-Ampofo, Stella and Zappacosta, Antony and Kuzu, Rıdvan Salih and Schauer, Peter and Willberg, Martin and Zhu, Xiao Xiang},
  journal={IEEE Geoscience and Remote Sensing Letters}, 
  title={SmallMinesDS: A Multi-Modal Dataset for Mapping Artisanal and Small-Scale Gold Mines}, 
  year={2025},
  volume={},
  number={},
  pages={1-1},
  keywords={Data mining;Foundation models;Gold;Satellite images;Sentinel-1;Optical sensors;Laser radar;Biological system modeling;Spaceborne radar;Radar imaging;Earth Observation;Semantic Segmentation;Mining;Machine Learning;Foundation Models},
  doi={10.1109/LGRS.2025.3566356}}

@ARTICaLE{id,
  author={Ofori-Ampofo, Stella and Zappacosta, Antony and Kuzu, Rıdvan Salih and Schauer, Peter and Willberg, Martin and Zhu, Xiao Xiang},
  journal={}, 
  title={Cocoa under Threat: Assessing the Impact of Artisanal Gold Mining on Cocoa Landscapes using Remote Sensing and Foundation Model}, 
  year={2025},
  volume={},
  number={},
  pages={},
  keywords={},
  doi={00.000/000000}}