NEE观测误差分布类型对陆地生态系统机理模型参数估计的影响——以长白山温带阔叶红松林为例

基于观测数据的陆地生态系统模型参数估计有助于提高模型的模拟和预测能力,降低模拟不确定性.在已有参数估计研究中,涡度相关技术测定的净生态系统碳交换量(NEE)数据的随机误差通常被假设为服从零均值的正态分布.然而近年来已有研究表明NEE数据的随机误差更服从双指数分布.为探讨NEE观测误差分布类型的不同选择对陆地生态系统机理模型参数估计以及碳通量模拟结果造成的差异,以长白山温带阔叶红松林为研究区域,采用马尔可夫链-蒙特卡罗方法,利用2003～2005年测定的NEE数据对陆地生态系统机理模型CEVSA2的敏感参数进行估计,对比分析了两种误差分布类型(正态分布和双指数分布)的参数估计结果以及碳通量模拟的差异.结果表明,基于正态观测误差模拟的总初级生产力和生态系统呼吸的年总量分别比基于双指数观测误差的模拟结果高61～86 g C m-2 a-1和107～116 g C m-2 a-1,导致前者模拟的NEE年总量较后者低29～47 g C m-2 a-1,特别在生长旺季期间有明显低估.在参数估计研究中,不能忽略观测误差的分布类型以及相应的目标函数的选择,它们的不合理设置可能对参数估计以及模拟结果产生较大影响.

Influences of error distributions of net ecosystem exchange on parameter estimation of a process-based terrestrial model

Accuracy of model predictions can be improved by parameter estimation from measurements. It was assumed that measurement errors of net ecosystem exchange of CO2 (NEE) by the eddy covariance technique follow a normal distribution. However, recent studies have showed that errors in eddy covariance measurements closely follow a double exponential rather than a normal distribution. In this paper, we compared effects of different distributions of measurement errors of NEE data on estimation of parameters and carbon fluxes components. Daily NEE measurements from 2003 to 2005 at the Changbaishan forest site were assimilated into a process-based terrestrial ecosystem model. The Markov Chain Monte Carlo method was used to derive the probability density functions of estimated parameters. Our results showed the modeled annual total gross primary production (GPP) and ecosystem respiration (Re) using the normal error distribution were higher than those using the double exponential distribution by 61-86 g C m-2 a-1 and 107-116 g C m-2 a-1, respectively. As a result, modeled annual sum of NEE under an assumption of the normal error distribution was lower by 29-47 g C m-2 a-1 than that under the double exponential error distribution. Especially, the modeled daily NEE based on the normal distribution underestimated the strong carbon sink in Changbaishan forest during the growing seasons. We concluded that types of measurement error distributions and corresponding cost functions can substantially influence parameter estimation and estimated carbon fluxes with data assimilation.