Scholarlink

<p>Sustainable drought risk management is essential for ensuring water security in arid regions, where climate change and land-use transitions intensify stress on fragile ecosystems. This study develops a geospatial artificial intelligence (GeoAI) framework that integrates climate model evaluation, remote sensing, deep learning and hydroclimatic assessment to support resilience planning in the Eastern Province of Saudi Arabia. Six CMIP6 climate models were benchmarked against ERA5-Land, with ACCESS-ESM1-5 demonstrating superior performance. A hybrid deep learning model combining artificial neural networks (ANN), long short-term memory (LSTM) and gated recurrent units (GRU) accurately predicted land surface temperature (LST), potential evapotranspiration (PET) and actual evapotranspiration (ET) (R² &gt; 0.94). From multiple climatic drivers (temperature, humidity, wind, precipitation, radiation), three indicators were derived: climatic water availability (CWA), crop water demand (CWD) and crop water stress index (CWSI). Sentinel-2 Dynamic World land-use/land-cover (LULC) data were incorporated to represent baseline and future conditions. Seasonal CWA risk maps and a composite drought index (CDI), integrating CWD and CWSI, were produced under three SSP scenarios (SSP1-2.6, SSP2-4.5, SSP5-8.5) across four-time horizons (near future, mid-century, far future, end-century). Results highlight intensifying drought severity under SSP5-8.5, especially in summer and late-century periods. Spatial agreement analysis (Kappa up to 0.69; Pearson <i>r</i> &gt; 0.9) confirmed strong consistency between water availability and drought classifications. The proposed framework offers a robust and scalable tool for anticipating drought risk and supporting adaptive land and water management in water-scarce environments.</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

Future water stress in arid landscapes projected with GeoAI

  • 期刊论文

  • 发表日期：

    2026年1月10日

查看全文
  • Mahfuzur Rahman,
  • Md Anuwer Hossain,
  • Mahmudur Rahman,
  • Ahadul Islam Patwary,
  • Md Masudur Rahman,
  • Mohammed Benaafi,
  • Isam H. Aljundi

摘要

Sustainable drought risk management is essential for ensuring water security in arid regions, where climate change and land-use transitions intensify stress on fragile ecosystems. This study develops a geospatial artificial intelligence (GeoAI) framework that integrates climate model evaluation, remote sensing, deep learning and hydroclimatic assessment to support resilience planning in the Eastern Province of Saudi Arabia. Six CMIP6 climate models were benchmarked against ERA5-Land, with ACCESS-ESM1-5 demonstrating superior performance. A hybrid deep learning model combining artificial neural networks (ANN), long short-term memory (LSTM) and gated recurrent units (GRU) accurately predicted land surface temperature (LST), potential evapotranspiration (PET) and actual evapotranspiration (ET) (R² > 0.94). From multiple climatic drivers (temperature, humidity, wind, precipitation, radiation), three indicators were derived: climatic water availability (CWA), crop water demand (CWD) and crop water stress index (CWSI). Sentinel-2 Dynamic World land-use/land-cover (LULC) data were incorporated to represent baseline and future conditions. Seasonal CWA risk maps and a composite drought index (CDI), integrating CWD and CWSI, were produced under three SSP scenarios (SSP1-2.6, SSP2-4.5, SSP5-8.5) across four-time horizons (near future, mid-century, far future, end-century). Results highlight intensifying drought severity under SSP5-8.5, especially in summer and late-century periods. Spatial agreement analysis (Kappa up to 0.69; Pearson r > 0.9) confirmed strong consistency between water availability and drought classifications. The proposed framework offers a robust and scalable tool for anticipating drought risk and supporting adaptive land and water management in water-scarce environments.