LAUSR

This study assesses the performance of temperature extremes over China in two regional climate models (RCMs), RegCM4 and WRF, driven by the ECMWF’s 20th century reanalysis. Based on the advice of the Expert Team on Climate Change Detection and Indices (ETCCDI), 12 extreme temperature indices (i.e., TXx, TXn, TNx, TNn, TX90p, TN90p, TX10p, TN10p WSDI, ID, FD, and CSDI) are derived from the simulations of two RCMs and compared with those from the daily station-based observational data for the period 1981–2010. Overall, the two RCMs demonstrate satisfactory capability in representing the spatiotemporal distribution of the extreme indices over most regions. RegCM performs better than WRF in reproducing the mean temperature extremes, especially over the Tibetan Plateau (TP). Moreover, both models capture well the decreasing trends in ID, FD, CSDI, TX10p, and TN10p, and the increasing trends in TXx, TXn, TNx, TNn, WSDI, TX90p, and TN90p, over China. Compared with observation, RegCM tends to underestimate the trends of temperature extremes, while WRF tends to overestimate them over the TP. For instance, the linear trends of TXx over the TP from observation, RegCM, and WRF are 0.53°C (10 yr)−1, 0.44°C (10 yr)−1, and 0.75°C (10 yr)−1, respectively. However, WRF performs better than RegCM in reproducing the interannual variability of the extreme-temperature indices. Our findings are helpful towards improving our understanding of the physical realism of RCMs in terms of different time scales, thus enabling us in future work to address the sources of model biases.摘 要以欧洲中心的二十世纪再分析资料为初始场和侧边界场, 利用两个区域气候模式—RegCM4.5 和 WRF3.6—对东亚地区的气候进行了水平分辨率为 50km 的动力降尺度模拟. 本文评估了两个模式在中国的极端气温模拟性能. 基于气候变化检测和指数专家组的建议, 对比分析了 1981–2010 年两个模式模拟的12个极端气温指数 (即TXx, TXn, TNx, TNn, TX90p, TN90p, TX10p, TN10p WSDI, ID, FD, 和CSDI) 及对应的观测资料. 总体而言, 在中国大多数地区, 两个区域气候模式在表征极端气温指数的时空分布特征上有令人满意的表征能力. RegCM模式再现极端气温的平均气候态上性能优于 WRF 模式, 尤其是在青藏高原地区. 此外, 两个模式都很好地抓住了 1981–2010 年间 ID, FD, CSDI, TX10p 和 TN10p 的下降趋势, 以及TXx, TXn, TNx, TNn, WSDI, TX90p 和 TN90p 的上升趋势. 在青藏高原地区, 与观测资料相比表明, RegCM 模式低估了极端气温的趋势, 而 WRF 模式高估了极端气温的趋势. 例如, 在青藏高原地区中TXx的线性趋势在观测资料, RegCM和WRF模式分别是 0.53ºC (10 yr)−1, 0.44ºC (10 yr)−1 和 0.75ºC (10 yr)−1. 在模拟极端气温的年际变率方面, WRF模式的表现优于 RegCM 模式. 我们的发现有助于提高对区域气候模式在不同时间尺度上物理过程的理解, 从而使我们在今后的工作中追踪模式偏差的来源