Future climate projections based on multi-model ensemble approach are seen as more reliable, but not all models are equally performant at reproducing climate features at a regional scale. An optimal regional GCM ensemble was developed for Sakha (Yakutia) Republic based on error statistics and spatial correlation metrics. Historical Coupled Model Intercomparison Project, version 6 (CMIP6) simulations from 48 global climate models (GCMs) were used to evaluate model quality compared to mean annual air temperature (MAAT) reanalysis data for 1961–1990, 1971–2000 and 1981–2010 reference periods, and the MAAT change between 1961-1990 and 1981–2010, ΔT. The best-performing reanalysis, GHCN-CAMS, was validated using observational data. This five-member ensemble includes CESM2-WACCM, CMCC-ESM2, CNRM-CM6-1-HR, INM-CM5-0, MPI-ESM1-2-HR models, weighted by Pearson's coefficient of spatial correlation between observed and modeled ΔT fields. Model weighting based on spatial correlation metrics improved the performance of the developed multi-model regional ensemble, which can be used in projecting future climate under different climate change scenarios.

中文翻译：

---

欧亚大陆北部萨哈（雅库特）共和国 CMIP6 全球气候模型区域集合

基于多模型集合方法的未来气候预测被认为更可靠，但并非所有模型在再现区域尺度的气候特征方面都具有同样的性能。基于误差统计和空间相关性度量，为萨哈（雅库特）共和国开发了最佳区域 GCM 集成。使用来自 48 个全球气候模型 (GCM) 的历史耦合模型比对项目第 6 版 (CMIP6) 模拟与 1961-1990 年、1971-2000 年和 1981-2010 年参考年平均气温 (MAAT) 再分析数据相比来评估模型质量期间，以及 1961-1990 和 1981-2010 之间的 MAAT 变化，ΔT。性能最佳的再分析 GHCN-CAMS 使用观察数据进行了验证。这个五成员集成包括 CESM2-WACCM、CMCC-ESM2、CNRM-CM6-1-HR、INM-CM5-0、MPI-ESM1-2-HR 模型，按观测和建模 ΔT 场之间的 Pearson 空间相关系数进行加权。基于空间相关性指标的模型加权提高了所开发的多模型区域集合的性能，可用于预测不同气候变化情景下的未来气候。