Scholarlink

<p>Ecological zones in Ghana are critical for biodiversity and livelihoods and have undergone significant transformations due to climate variability and other factors. While prior studies have focused on Ghana’s ecosystems, a comprehensive analysis of long-term spatiotemporal changes across all six ecological zones, coastal savanna (CS), Guinea savanna (GS), Sudan savanna (SS), tropical rainforest (TRF), semi-deciduous forest (SDF), and transitional zone (TZ) remains limited. This study addresses this gap by integrating multispectral Landsat imagery (1985–2023) with three machine learning algorithms: maximum likelihood classification (MLC), support vector machine (SVM), and random forest (RF), to quantify historical trends, drivers, and ecological implications. Results revealed divergent trajectories: Sudan savanna (+ 2.38%) and Guinea savanna (+ 1.12%) expanded, while semi-deciduous forest (−1.15%), transitional zone (−0.78%), coastal savanna (−0.36%), and tropical rainforest (−1.23%) declined annually, respectively. The random forest classifier achieved the highest accuracy (<i>κ</i> = 0.89; producer accuracy = 90%), outperforming the support vector machine (<i>κ</i> = 0.85) and maximum likelihood classification (<i>κ</i> = 0.81), particularly in heterogeneous zones like the transitional zone and Sudan savanna. These shifts reflect dual pressures of climatic stressors (reduced rainfall, temperature fluctuations) and anthropogenic activities (deforestation, urbanisation, mining), with arid savanna conditions encroaching into forested regions, signalling climate-induced desertification. The study’s novel integration of machine learning with remote sensing provides a robust framework for monitoring ecological transitions in dynamic tropical landscapes. By bridging technological advancements with ecological management, this work offers actionable insights for policymakers to mitigate degradation, promote sustainable land use, and enhance resilience in Ghana and analogous regions facing similar environmental challenges.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Spatio-temporal analysis of Ghana’s ecological zones: integrating multispectral remote sensing and machine learning algorithms

  • 期刊论文

  • 发表日期：

    2026年3月26日

查看全文
  • Augustine O. K. N. Mensah,
  • Bi Shuoben,
  • Clement Kwang,
  • Emmanuel Yeboah,
  • Isaac Sarfo,
  • Ebenezer Nikoi,
  • Collins Oduro,
  • Prince Martin Gyekye,
  • Ishmeal Quist,
  • Princella Ofori Darko,
  • Clement Issaka Anaba,
  • Asante Vincent Antwi

摘要

Ecological zones in Ghana are critical for biodiversity and livelihoods and have undergone significant transformations due to climate variability and other factors. While prior studies have focused on Ghana’s ecosystems, a comprehensive analysis of long-term spatiotemporal changes across all six ecological zones, coastal savanna (CS), Guinea savanna (GS), Sudan savanna (SS), tropical rainforest (TRF), semi-deciduous forest (SDF), and transitional zone (TZ) remains limited. This study addresses this gap by integrating multispectral Landsat imagery (1985–2023) with three machine learning algorithms: maximum likelihood classification (MLC), support vector machine (SVM), and random forest (RF), to quantify historical trends, drivers, and ecological implications. Results revealed divergent trajectories: Sudan savanna (+ 2.38%) and Guinea savanna (+ 1.12%) expanded, while semi-deciduous forest (−1.15%), transitional zone (−0.78%), coastal savanna (−0.36%), and tropical rainforest (−1.23%) declined annually, respectively. The random forest classifier achieved the highest accuracy (κ = 0.89; producer accuracy = 90%), outperforming the support vector machine (κ = 0.85) and maximum likelihood classification (κ = 0.81), particularly in heterogeneous zones like the transitional zone and Sudan savanna. These shifts reflect dual pressures of climatic stressors (reduced rainfall, temperature fluctuations) and anthropogenic activities (deforestation, urbanisation, mining), with arid savanna conditions encroaching into forested regions, signalling climate-induced desertification. The study’s novel integration of machine learning with remote sensing provides a robust framework for monitoring ecological transitions in dynamic tropical landscapes. By bridging technological advancements with ecological management, this work offers actionable insights for policymakers to mitigate degradation, promote sustainable land use, and enhance resilience in Ghana and analogous regions facing similar environmental challenges.