青藏高原中东部地区土壤湿度模拟性能评估以及误差分析

土壤湿度不仅是地表水循环的重要组成部分,而且对天气和气候也有重要影响,它的模拟误差严重阻碍了人们对水循环的认知.本文首先评估了1°×1°水平分辨率的全球陆面数据同化产品(Global Land Data Assimilation System,GLDAS)对青藏高原中部那曲地区和东部玛曲地区土壤湿度的模拟性能;鉴于GLDAS较粗的分辨率无法精细描述分析区域土壤湿度空间分布特征,于是我们基于通用陆面过程模式(Community Land Surface Model,Version 4.5),开展了高分辨率0.1°×0.1°的模拟,并对高分辨率模拟土壤湿度误差的原因进行了深入分析.结果表明:(1)GLDAS陆面数据同化产品和高分辨率陆面模式模拟结果都可以反映出土壤湿度的季节变化特征,但在非冻结期均存在不同程度的干偏差,尤其是在玛曲地区;(2)对比观测和模拟的土壤湿度发现,观测数据表现出强烈的空间异质性,而模拟结果呈现出的是空间均一性.按照模拟误差进行归类分组,对比模拟性能优劣的两组站点发现,模式物理过程不是模拟性能差异的主要因素,而两组站点间地表特征参数中的土壤质地和地形参数,以及驱动数据均没有体现出空间异质性,这可能是土壤湿度模拟结果没有表现出空间异质性的原因.

Evaluation of soil-moisture simulations in the central and eastern Tibetan Plateau and error analysis

Soil moisture not only is an important component of the terrestrial water cycle, but also has a significant impact on weather and climate. The errors of soil moisture simulations can pose a serious effect on understanding of water cycle. This study firstly evaluates the performance of simulated soil moisture in the two soil moisture observation networks in the Tibetan Plateau by using the 1°×1° horizontal resolution Global Land Data Assimilation System (GLDAS). Because of its poor horizontal resolution, this system is unable to describe the fine spatial distribution characteristics of soil moisture in the network. Therefore, we conduct a simulation with a high horizontal resolution of 0.1°×0.1° based on the Community Land Surface Model (CLM, Version 4.5) to evaluate the performance of the simulation results and analyze the cause of errors in soil moisture simulation. Results show that: (1) GLDAS products and high-resolution CLM4.5 simulation can well reflect the seasonal variation of observed soil moisture, but all display underestimates in both of the two observation network regions during the unfrozen period, especially in the Maqu network region. (2) Comparing observation data with simulation results indicates that observation data have strong heterogeneity, but the simulation results do not. According to error statistics, we divide these sites into two groups and comparing better simulated sites with poor simulated sites. Results suggest that physical processes are not main reason for different simulation results between sites in the same network. While the land surface parameters and forcing data in the models can not reflect strong heterogeneity as observation data, which are probably the main reason for that the high-resolution simulation of soil moisture does not show strong heterogeneity.