Scholarlink

<p>Bias nonstationarity refers to the phenomenon where biases in climate model outputs vary over time rather than remaining consistent, thereby challenging the reliability of bias correction methods. This study employs the pseudo-reality approach—a way that uses one Global Climate Model (GCM) as a reference to estimate biases in other GCMs—to investigate the effects of bias nonstationarity on precipitation and temperature projections in Iran. Fourteen GCMs from the Coupled Model Intercomparison Project 6 (CMIP6) were analyzed using both univariate and multivariate bias correction methods. The results demonstrate that bias correction methods effectively improve climate model outputs when biases remain stationary. However, under nonstationary conditions, these methods introduce additional uncertainties, particularly for precipitation. Bias nonstationarity significantly impacts future precipitation projections. In the case of temperature, although relatively large biases are present, the smaller magnitude of bias changes across two periods compared to the climate change signal reduces the critical impact of bias nonstationarity on future temperature projections. The uncertainty analysis highlights that in the present, model uncertainty and internal variability (IV) are the primary sources of uncertainty for both variables. As the 21st century progresses, IV’s contribution to temperature uncertainty decreases dramatically from 30% to less than 1%, while scenario uncertainty grows to 55%, becoming the dominant source. For precipitation, IV declines from 40 to 10%, while model uncertainty increases substantially from 40 to 70%, emerging as the leading source of uncertainty. Despite the growth of scenario uncertainty over time, its contribution remains relatively minor compared to model uncertainty and IV. Furthermore, incorporating observational data into the pseudo-reality approach, referred to as a semi-reality approach, reveals that using real-world measurements emphasizes the critical impact of bias nonstationarity on bias correction outcomes. This underscores the importance of addressing bias nonstationarity in future climate modeling practices. The development of more robust methodologies for climate adaptation enables better decision-making strategies in the face of uncertainty.</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

Assessing temperature and precipitation bias nonstationarity of CMIP6 global climate models over Iran

  • 期刊论文

  • 发表日期：

    2025年1月10日

查看全文
  • Narges Azad,
  • Azadeh Ahmadi

摘要

Bias nonstationarity refers to the phenomenon where biases in climate model outputs vary over time rather than remaining consistent, thereby challenging the reliability of bias correction methods. This study employs the pseudo-reality approach—a way that uses one Global Climate Model (GCM) as a reference to estimate biases in other GCMs—to investigate the effects of bias nonstationarity on precipitation and temperature projections in Iran. Fourteen GCMs from the Coupled Model Intercomparison Project 6 (CMIP6) were analyzed using both univariate and multivariate bias correction methods. The results demonstrate that bias correction methods effectively improve climate model outputs when biases remain stationary. However, under nonstationary conditions, these methods introduce additional uncertainties, particularly for precipitation. Bias nonstationarity significantly impacts future precipitation projections. In the case of temperature, although relatively large biases are present, the smaller magnitude of bias changes across two periods compared to the climate change signal reduces the critical impact of bias nonstationarity on future temperature projections. The uncertainty analysis highlights that in the present, model uncertainty and internal variability (IV) are the primary sources of uncertainty for both variables. As the 21st century progresses, IV’s contribution to temperature uncertainty decreases dramatically from 30% to less than 1%, while scenario uncertainty grows to 55%, becoming the dominant source. For precipitation, IV declines from 40 to 10%, while model uncertainty increases substantially from 40 to 70%, emerging as the leading source of uncertainty. Despite the growth of scenario uncertainty over time, its contribution remains relatively minor compared to model uncertainty and IV. Furthermore, incorporating observational data into the pseudo-reality approach, referred to as a semi-reality approach, reveals that using real-world measurements emphasizes the critical impact of bias nonstationarity on bias correction outcomes. This underscores the importance of addressing bias nonstationarity in future climate modeling practices. The development of more robust methodologies for climate adaptation enables better decision-making strategies in the face of uncertainty.