| 引用本文: | 张婷,王粼昊,李建柱,冯平.2025.全球气候模式在区域尺度上的适用性及其应用研究进展[J].地球环境学报,16(2):107-121 |
| ZHANG Ting,WANG Linhao,LI Jianzhu,FENG Ping.2025.Advance on applicability and application of global climate model at regional scale[J].Journal of Earth Environment,16(2):107-121 |
|
| 摘要: |
| 气候变化对流域的水文循环和水资源分布产生一定的影响，选择合适的全球气候模式（global climate model，GCM），通过降尺度处理，获得小尺度、高分辨率、高精度的气候信息，是研究区域尺度上未来气候变化及其水文响应的前提。基于上述背景，通过对当前研究成果进行全面回顾，进而归纳气候模式适用性评估、降尺度技术和偏差校正研究中所涉及的方法，总结各类方法的优点和面临的挑战。从历史时期法、多模式集合法和模式独立性方法三个角度系统阐述气候模式适用性评估方法；从动力降尺度、统计降尺度和动力-统计降尺度三方面对降尺度方法进行梳理和评述；介绍研究中常见的偏差校正方法及其应用情况；最后，综合分析气候模式适用性评估方法、降尺度方法和偏差校正方法的发展趋势。研究将为全球气候模式在区域尺度上的适用性评估及其应用研究提供方法和思路指导。 |
| 关键词: 气候变化 全球气候模式 评估方法 降尺度方法 偏差校正方法 |
| DOI：10.7515/JEE221032 |
| CSTR：32259.14.JEE221032 |
| 分类号: |
| 基金项目:国家自然科学基金项目（52079086）；天津市自然科学基金项目（20JCQNJC01960） |
| 英文基金项目:National Natural Science Foundation of China (52079086); Natural Science Foundation of Tianjin (20JCQNJC01960) |
|
| Advance on applicability and application of global climate model at regional scale |
|
ZHANG Ting*, WANG Linhao, LI Jianzhu, FENG Ping
|
|
State Key Laboratory of Hydraulic Engineering Intelligent Construction and Operation, Tianjin University, Tianjin 300350, China
|
| Abstract: |
| Background, aim, and scope Global warming directly affects the climate system, intensifies the water cycle, increases the occurrence probability of extreme hydrological events, and changes the water balance in basins. Selecting the best global climate model (GCM) to obtain small-scale, high-resolution, and high-precision climate information through downscaling is a prerequisite for studying future climate change and hydrological responses at the regional scale. To clarify the general approach for applying GCMs on a regional scale, this study provides a systematic summary of the methods used for GCM suitability assessment, downscaling, and bias correction. Materials and methods The historical period technique, multimode ensemble averaging approach, and model independence method were employed to comprehensively examine methodologies for evaluating the applicability of climate models. Dynamic downscaling and statistical downscaling were separated into different downscaling techniques. Common deviation correction techniques were introduced, along with their applications. Results GCMs can be used to simulate the evolution of climate states under specified forcings and constitute the primary tool for predicting future climate changes and their hydrological responses. The historical period method is straightforward but limited by the correlation between historical and future signals; the adoption of multimodel ensemble approaches can reduce uncertainties, but these approaches are insensitive to extreme values, while quantifying model independence remains challenging. Statistical downscaling is convenient yet lacks a physical basis, whereas dynamical downscaling incorporates physical principles but incurs high computational costs. Although effective, bias correction methods exhibit limited precision. Discussion The core assumption of the historical period method: the historical performance determines the future accuracy, is fundamentally undermined by the spatiotemporal variability in climate behavior, forcing single-variable methodologies to neglect critical interdependencies. Multimodel ensemble approaches, while reducing uncertainty through averaging, fail to address systematically correlated biases across models, thus exacerbating errors in extreme-value projections. Model independence remains elusive because of the use of nonstandardized evaluation metrics and opaqueness in lineage tracing, which limits ensemble diversification. In the physics-based framework of dynamical downscaling, biases are propagated from coarse-resolution drivers owing to its unidirectional coupling mechanism. The computational efficiency of statistical downscaling fails to compensate for the absence of elevation-precipitation interaction modeling and fragmented spatial outputs in topographically complex regions. Bias correction methods, while preserving macroscale signals, fail to account for evolving variable relationships under shifting boundary conditions. Conclusions Quantitative statistical methods constitute the primary method for assessing the applicability of current climate models at the regional scale. The bias correction approach and downscaling method must be applied in accordance with a thorough analysis of regional conditions and climate factor characteristics. Recommendations and perspectives Future studies should be performed from the following perspectives: improving the quality of observational data; combining statistics and dynamics; and comprehensively understanding the physical, chemical, biological, and other processes in climate models. |
| Key words: climate change global climate model evaluation methods downscaling methods bias correction |
