Scholarlink

<p>Reliable fine-scale hail projections are needed for robust risk assessments, yet future trends remain uncertain and contradictory across studies. Using 11-year pan-European convection-permitting climate simulations for the present and a + 3 °C pseudo-global-warming climate, we compare an online hail-growth diagnostic (HAILCAST) with an offline machine learning model (XGBoost) trained on ERA5 hail environments. Under present conditions, both approaches provide a plausible European hail climatology. However, XGBoost predicts widespread hail suppression in a warmer climate, mainly driven by increasing freezing-level heights outside the training distribution. HAILCAST instead simulates how enhanced storm updrafts can sustain hail growth despite a warmer atmosphere, projecting increases over central-eastern Europe and larger hailstones. The results show that data-driven approaches alone should be used with caution under global warming, because the relationships learned under present-day conditions are not climate-change invariant, underscoring the need for physically based hail representations in convection-permitting models to robustly assess future convective hazards.</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

Divergent European hail projections from machine learning and physically based models under global warming

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月8日

查看全文
  • Marco Chericoni,
  • Giorgia Fosser,
  • Alessandro Anav,
  • Iris Thurnherr,
  • Andreas F. Prein

摘要

Reliable fine-scale hail projections are needed for robust risk assessments, yet future trends remain uncertain and contradictory across studies. Using 11-year pan-European convection-permitting climate simulations for the present and a + 3 °C pseudo-global-warming climate, we compare an online hail-growth diagnostic (HAILCAST) with an offline machine learning model (XGBoost) trained on ERA5 hail environments. Under present conditions, both approaches provide a plausible European hail climatology. However, XGBoost predicts widespread hail suppression in a warmer climate, mainly driven by increasing freezing-level heights outside the training distribution. HAILCAST instead simulates how enhanced storm updrafts can sustain hail growth despite a warmer atmosphere, projecting increases over central-eastern Europe and larger hailstones. The results show that data-driven approaches alone should be used with caution under global warming, because the relationships learned under present-day conditions are not climate-change invariant, underscoring the need for physically based hail representations in convection-permitting models to robustly assess future convective hazards.