Scholarlink

<p>The process of industrialization of arid areas can rapidly surpass the pace of environmental monitoring, which leaves a gap in the knowledge of long-term processes of soil degradation. Although conventional monitoring usually involves the use of a static-based evaluation that only records the existing changes in time, this research establishes a dynamic spatio-temporal model to rebuild 30&#xa0;years of soil dynamics. To support the backcasting of soil conditions in 1986, 1999 and 2010, we combined a Random Forest (RF) model with the Iranian Model of Desertification Potential Assessment (IMDPA). All the data in satellites were strictly radiometrically normalized to provide spectral consistency in Landsat 5, 7 and 8 sensors. The RF model, which has been trained on 201 soil samples in 2016, was able to predict important chemical makers such as Electrical Conductivity (EC), Sodium Adsorption Ratio (SAR), soil pH and the Soil Quality Index (SQI). Without the historical ground-truth data, the accuracy of the temporal reconstructions was evaluated using the spatial uncertainty mapping and by cross-checking with the local history. We find that the natural hydrological processes cause groundwater salinity, whereas closeness to mines and cities increases the erosion of soil quality. Spatiotemporal reconstruction showed that the Very High desertification class increased 60-fold from 29&#xa0;ha in 1986 up to 1766&#xa0;ha in 2016, mostly concentrated around the Yazd-Ardakan industrial belt. This spatial association between urban growth and land degradation has offered a quantitative diagnostic floor in specific land management and selective reclamation in perilous aridarious settings.</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

Spatiotemporal mapping of chemical soil degradation in Iranian arid lands using legacy soil data and hybrid RF–IMDPA

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月19日

查看全文
  • Hassan Fathizad,
  • Mohammad Ali Hakimzadeh Ardakani

摘要

The process of industrialization of arid areas can rapidly surpass the pace of environmental monitoring, which leaves a gap in the knowledge of long-term processes of soil degradation. Although conventional monitoring usually involves the use of a static-based evaluation that only records the existing changes in time, this research establishes a dynamic spatio-temporal model to rebuild 30 years of soil dynamics. To support the backcasting of soil conditions in 1986, 1999 and 2010, we combined a Random Forest (RF) model with the Iranian Model of Desertification Potential Assessment (IMDPA). All the data in satellites were strictly radiometrically normalized to provide spectral consistency in Landsat 5, 7 and 8 sensors. The RF model, which has been trained on 201 soil samples in 2016, was able to predict important chemical makers such as Electrical Conductivity (EC), Sodium Adsorption Ratio (SAR), soil pH and the Soil Quality Index (SQI). Without the historical ground-truth data, the accuracy of the temporal reconstructions was evaluated using the spatial uncertainty mapping and by cross-checking with the local history. We find that the natural hydrological processes cause groundwater salinity, whereas closeness to mines and cities increases the erosion of soil quality. Spatiotemporal reconstruction showed that the Very High desertification class increased 60-fold from 29 ha in 1986 up to 1766 ha in 2016, mostly concentrated around the Yazd-Ardakan industrial belt. This spatial association between urban growth and land degradation has offered a quantitative diagnostic floor in specific land management and selective reclamation in perilous aridarious settings.