CAJ | 학술논문

　　基于五变量草原生态系统理论模式，应用与参数有关的条件非线性最优扰动(CNOP-P)方法，探讨了由参数不确定性导致的草原生态系统模式模拟结果的不确定性问题。参数的不确定性可能来源于观测和（或）对物理过程描述等的不确定性。选取了五变量草原生态系统模式中具有物理意义的32个模式参数进行数值试验。试验结果表明，对所考察的32个模式参数，在一定的不确定性和给定的优化时刻范围内，单独优化每个参数所得CNOP-Ps的联合模态与同时优化32个参数所得CNOP-P的模态并不相同。比较了上述两类参数误差以及随机参数误差对草原生态系统模拟的差异。随机参数误差与上述优化方法所得参数误差的不确定性范围大小相同。数值结果表明，同时优化32个参数所得 CNOP-P 类型参数误差使得草原生态系统模拟的不确定性程度最大。这种影响表现在使得草原生态系统转变为沙漠生态系统，或者使得草原生态系统转变为具有更多生草量的草原生态系统。上述数值结果不依赖于优化时间和参数不确定性程度的大小。这些数值结果建议我们应当考虑多参数的非线性相互作用来研究草原生态系统模式模拟的不确定性问题，并且揭示出CNOP-P方法是讨论上述问题的一个有用的工具。
　　기우오변량초원생태계통이론모식，응용여삼수유관적조건비선성최우우동(CNOP-P)방법，탐토료유삼수불학정성도치적초원생태계통모식모의결과적불학정성문제。삼수적불학정성가능래원우관측화（혹）대물리과정묘술등적불학정성。선취료오변량초원생태계통모식중구유물리의의적32개모식삼수진행수치시험。시험결과표명，대소고찰적32개모식삼수，재일정적불학정성화급정적우화시각범위내，단독우화매개삼수소득CNOP-Ps적연합모태여동시우화32개삼수소득CNOP-P적모태병불상동。비교료상술량류삼수오차이급수궤삼수오차대초원생태계통모의적차이。수궤삼수오차여상술우화방법소득삼수오차적불학정성범위대소상동。수치결과표명，동시우화32개삼수소득 CNOP-P 류형삼수오차사득초원생태계통모의적불학정성정도최대。저충영향표현재사득초원생태계통전변위사막생태계통，혹자사득초원생태계통전변위구유경다생초량적초원생태계통。상술수치결과불의뢰우우화시간화삼수불학정성정도적대소。저사수치결과건의아문응당고필다삼수적비선성상호작용래연구초원생태계통모식모의적불학정성문제，병차게시출CNOP-P방법시토론상술문제적일개유용적공구。
The uncertainties in grassland ecosystem simulations caused by uncertainties in the parameters were studied using a theoretical five-variable grassland ecosystem model and a conditional nonlinear optimal perturbation (CNOP-P) method. Uncertainties in the parameters may originate in uncertainties in the observations and/or descriptions of the physical processes associated with the parameter, amongst other things. 32 model parameters that have physical meanings in the five-variable grassland ecosystem model were selected for use in numerical experiments. The results showed that when these parameters had the same degree of uncertainty, and the same optimization time, the combination of CNOP-Ps optimized for each parameter was different from the CNOP-P optimized for all 32 model parameters. The authors compared the grassland ecosystem simulations with the two types of parameter errors described above and with random parameter errors with the same degree of uncertainty as the optimized parameter errors. It was concluded that the CNOP-P for the 32 model parameters optimized at the same time led to the maximum uncertainty in the grassland ecosystem simulation, which was that the grassland ecosystem was either transformed into a desert ecosystem or another grassland ecosystem with more biomass. These results were independent of the size of the parameter uncertainties and the optimization time, and they show that nonlinear interactions between several parameters in the model are important to the uncertainties in the grassland ecosystem simulation. The results also imply that the CNOP-P method is a useful tool for assessing uncertainties in the grassland ecosystem simulation.