Scholarlink

<p>Gully erosion represents a severe geomorphic threat to tropical landscapes, yet multi-temporal assessments of its expansion and comparative susceptibility modelling remain under-researched. This study implements an integrated geomatics pipeline to quantify gully dynamics and model erosion susceptibility in two contrasting formations in Anambra State, Nigerian: the gully-prone Nanka sands formation of Idemili and the stable Imo clay formations of Awka area. Utilising the high-resolution Google Earth imagery of years 2000 and 2025, gully inventories were manually digitised, revealing a significant intensification of land degradation. In Idemili, gully frequency rose by 69%, with a 181.5% increase in surface area, while Awka experienced a 169.6% increase in surface area. To isolate the environmental drivers of this instability, an eleven-variable predictor suite comprising geomorphometric and pedological indices was integrated via Google Earth Engine and SoilGrids v2.0. Susceptibility was then modelled through a comparative framework, leveraging the non-linear predictive power of Random Forest (RF) alongside the inferential precision of Binary Logistic Regression (BLR). Performance metrics demonstrated that RF significantly outperformed BLR, achieving a superior recall of 0.9057 and an Area Under Curve (AUC) of 0.825. Variable importance analysis identified elevation, Topographic Wetness Index (TWI), and slope as primary topographic drivers, while high sand-to-clay ratios in Idemili served as critical pedological triggers. Susceptibility mapping confirmed that 29.43% and 28.65% of the Ameki Group falls within Moderate to High-risk zones respectively. On the other hand, for Awka the moderate to high-risk zones is 28.25% and 10.38% respectively. These results validate a transferable geomatics framework for erosion monitoring and underscore the necessity of integrating catchment-scale hydrology with soil-mineralogical properties to effectively reduce disaster risk.</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

Integrating machine learning and geomatics in modelling gully erosion susceptibility in Awka and Idemili areas of Anambra state, Nigeria

  • 期刊论文

  • 发表日期：

    2026年6月22日

查看全文
  • Nelson Onyebuchi Nwobi

摘要

Gully erosion represents a severe geomorphic threat to tropical landscapes, yet multi-temporal assessments of its expansion and comparative susceptibility modelling remain under-researched. This study implements an integrated geomatics pipeline to quantify gully dynamics and model erosion susceptibility in two contrasting formations in Anambra State, Nigerian: the gully-prone Nanka sands formation of Idemili and the stable Imo clay formations of Awka area. Utilising the high-resolution Google Earth imagery of years 2000 and 2025, gully inventories were manually digitised, revealing a significant intensification of land degradation. In Idemili, gully frequency rose by 69%, with a 181.5% increase in surface area, while Awka experienced a 169.6% increase in surface area. To isolate the environmental drivers of this instability, an eleven-variable predictor suite comprising geomorphometric and pedological indices was integrated via Google Earth Engine and SoilGrids v2.0. Susceptibility was then modelled through a comparative framework, leveraging the non-linear predictive power of Random Forest (RF) alongside the inferential precision of Binary Logistic Regression (BLR). Performance metrics demonstrated that RF significantly outperformed BLR, achieving a superior recall of 0.9057 and an Area Under Curve (AUC) of 0.825. Variable importance analysis identified elevation, Topographic Wetness Index (TWI), and slope as primary topographic drivers, while high sand-to-clay ratios in Idemili served as critical pedological triggers. Susceptibility mapping confirmed that 29.43% and 28.65% of the Ameki Group falls within Moderate to High-risk zones respectively. On the other hand, for Awka the moderate to high-risk zones is 28.25% and 10.38% respectively. These results validate a transferable geomatics framework for erosion monitoring and underscore the necessity of integrating catchment-scale hydrology with soil-mineralogical properties to effectively reduce disaster risk.