数值模式中物理参数的不确定性是导致陆地生态系统模拟和预测不确定性的一个主要因素。减少数值模式中物理参数的误差是提高陆地生态系统模拟能力和预报技巧的有效途径之一。陆地生态系统模式中含有大量的由直接或间接观测可以确定的物理参数(Parameters determined by direct and (or) indirect observation, PDOs)，通过观测减少所有这些物理参数的误差将花费巨大的代价。Mu（2013）提出了减小该类物理参数误差的目标观测思想。本项目首先利用基于条件非线性最优参数扰动（CNOP-P）的识别敏感物理参数的新方法，在PDOs中寻找对陆地生态系统模拟和预测影响最重要的一小部分参数，进而通过目标观测减少它们的误差；探讨减少这类物理参数误差对于陆地生态系统模拟和预测的收益程度及其物理机制，为通过目标观测提高陆地生态系统模式的模拟与预报技巧提供科学方法与支撑。

One main factor of causing uncertainty of numerical simulation and prediction is uncertainties of physical parameters in numerical models. It is effective tool to improve simulation ability and prediction skill of terrestrial ecosystem to reduce the errors of physical parameters in numerical models. There are lots of physical parameters determined by direct and (or) indirect observation (PDOs) in numerical models related to terrestrial ecosystem. It will cost tremendous computational cost to reduce all errors of the PDOs with observation data. The idea of targeted observation to reduce the errors of the PDOs with observation data was proposed by Mu (2013). In this project, a small part of relative sensitive and important PDOs among all PDOs are determined with the new approach of based on the conditional nonlinear optimal parameter perturbation (CNOP-P) to identify the sensitive and important parameters. The benefit and its physical mechanism are analyzed to numerical simulation and prediction of terrestrial ecosystem through reducing the errors of relative sensitive and important PDOs. The study supplies scientific method and support to improve the simulation ability and prediction skill about the terrestrial ecosystem through the idea of targeted observation.