Scholarlink

<p>The spatiotemporal patterns of Land Surface Temperature (LST) across Iceland from 2001 to 2023 were analyzed using 1&#xa0;km‑resolution MODIS MOD11A2 8‑day composites, and an Artificial Neural Network (ANN) was applied to project LST for 2035. Daytime (LSTd) and nighttime (LSTn) temperatures were examined across the country’s eight major climate zones. Results reveal pronounced seasonal and elevational gradients: winter LSTd predominantly ranges from − 10&#xa0;°C to 0&#xa0;°C, rising to 10–20&#xa0;°C in summer, while winter LSTn frequently falls between − 20&#xa0;°C and − 5&#xa0;°C, especially in glaciated interiors, with reduced diurnal variability due to persistent snow cover. Trend analysis shows that over the 22-year period, approximately 53% of Iceland’s surface warmed by day, and 69% by night, particularly in eastern and northeastern regions, while only January (LSTn) and December (both LSTd &amp; LSTn) exhibited significant cooling (<i>p</i> ≤ 0.05). The ANN model, trained on 2023 data, achieved high predictive accuracy (R² &gt;0.92 and RMSE of 0.06&#xa0;°C for LSTn and &lt; 0.2&#xa0;°C for LSTd). Projections for 2035 indicate up to + 4&#xa0;°C daytime warming in central glaciated zones and nighttime cooling to -10&#xa0;°C. These asymmetric thermal shifts suggest heightened glacier instability, periglacial erosion, and ecological stress. These findings underscore the effectiveness of machine learning in high‑latitude climate analysis and highlight the urgent need of data‑driven adaptation strategies.</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 dynamics and ANN-based projection of daytime and nighttime land surface temperature in Iceland (2001–2035)

  • 期刊论文

  • 发表日期：

    2025年7月19日

查看全文
  • Haraldur Olafsson,
  • Iman Rousta,
  • Marjan Dalvi,
  • Jaromir Krzyszczak

摘要

The spatiotemporal patterns of Land Surface Temperature (LST) across Iceland from 2001 to 2023 were analyzed using 1 km‑resolution MODIS MOD11A2 8‑day composites, and an Artificial Neural Network (ANN) was applied to project LST for 2035. Daytime (LSTd) and nighttime (LSTn) temperatures were examined across the country’s eight major climate zones. Results reveal pronounced seasonal and elevational gradients: winter LSTd predominantly ranges from − 10 °C to 0 °C, rising to 10–20 °C in summer, while winter LSTn frequently falls between − 20 °C and − 5 °C, especially in glaciated interiors, with reduced diurnal variability due to persistent snow cover. Trend analysis shows that over the 22-year period, approximately 53% of Iceland’s surface warmed by day, and 69% by night, particularly in eastern and northeastern regions, while only January (LSTn) and December (both LSTd & LSTn) exhibited significant cooling (p ≤ 0.05). The ANN model, trained on 2023 data, achieved high predictive accuracy (R² >0.92 and RMSE of 0.06 °C for LSTn and < 0.2 °C for LSTd). Projections for 2035 indicate up to + 4 °C daytime warming in central glaciated zones and nighttime cooling to -10 °C. These asymmetric thermal shifts suggest heightened glacier instability, periglacial erosion, and ecological stress. These findings underscore the effectiveness of machine learning in high‑latitude climate analysis and highlight the urgent need of data‑driven adaptation strategies.