Scholarlink

Abstract <p>Quantile Random Forest (QRF) algorithm was applied to produce a Digital Soil Map of Soil Organic Carbon (SOC) across the eastern slopes of Mount Kenya. Soil samples from 70 locations designed by the Conditional Latin Hypercube Sampling (cLHS) method were collected. Environmental covariates were acquired and derived from Google Earth Engine included DEM, terrain derivations, and Landsat bands and indices. The QRF model was tuned with a search grid strategy and validated using 15-fold cross-validation with 100 repetitions plus external validation (Test set), achieving a Root Mean Squared Error (RMSE) of 1.11 and coefficient of determination (<i>R</i><sup>2</sup>) of 0.41 in average for the cross-validation, and RMSE of 1.31 and <i>R</i><sup>2</sup> of 0.5 for the Test. The SOC predictions correlated with elevation and other environmental covariates, showing higher SOC at higher altitudes. The study area embraces a range of World Reference Base for Soil Resources (WRB) Reference Soil Groups (RSGs) such as Nitisols, Ferralsols, Vertisols and Phaeozems.</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

Digital Soil Mapping of Soil Organic Carbon at Eastern Slopes of Mount Kenya

  • 期刊论文

  • 发表日期：

    2025年9月9日

查看全文
  • Y. A. Gelsleichter,
  • A. J. Toth,
  • E. Micheli,
  • E. Mutuma,
  • Amos Wawire,
  • A. Csorba

摘要

Abstract

Quantile Random Forest (QRF) algorithm was applied to produce a Digital Soil Map of Soil Organic Carbon (SOC) across the eastern slopes of Mount Kenya. Soil samples from 70 locations designed by the Conditional Latin Hypercube Sampling (cLHS) method were collected. Environmental covariates were acquired and derived from Google Earth Engine included DEM, terrain derivations, and Landsat bands and indices. The QRF model was tuned with a search grid strategy and validated using 15-fold cross-validation with 100 repetitions plus external validation (Test set), achieving a Root Mean Squared Error (RMSE) of 1.11 and coefficient of determination (R2) of 0.41 in average for the cross-validation, and RMSE of 1.31 and R2 of 0.5 for the Test. The SOC predictions correlated with elevation and other environmental covariates, showing higher SOC at higher altitudes. The study area embraces a range of World Reference Base for Soil Resources (WRB) Reference Soil Groups (RSGs) such as Nitisols, Ferralsols, Vertisols and Phaeozems.